diff --git "a/mtplx_build_metadata.json" "b/mtplx_build_metadata.json" --- "a/mtplx_build_metadata.json" +++ "b/mtplx_build_metadata.json" @@ -1,13528 +1,38 @@ { - "artifact_audit": { - "histogram": { - "bits": { - "8": 498 - }, - "bits_group_size": { - "8bit_g64": 498 - }, - "module_groups": { - "embed_tokens": 1, - "linear_attn": 240, - "lm_head": 1, - "mlp": 192, - "self_attn": 64 - } - }, - "model_dir": "/Users/youssof/Documents/MTPLX/models/Qwen3.6-27B-MTPLX-Flat8-Quality-TargetBase", - "passed": true, - "policy": "mtplx-flat8-quality", - "problems": [], - "protected_quantized_entries": [], - "quantization_entries": 498, - "required_bad": [], - "required_missing": [], - "sidecars": { - "chat_template.jinja": true, - "preprocessor_config.json": true, - "processor_config.json": true, - "tokenizer.json": true, - "tokenizer_config.json": true, - "video_preprocessor_config.json": true - } + "base": { + "layout": "official six-shard MLX VLM trunk", + "repo_id": "mlx-community/Qwen3.6-27B-8bit", + "revision": "c5a593c1475a746e43a543b0a02bd2b357e5745f" }, - "dtype": "bfloat16", - "mode": "build", - "output_path": "/Users/youssof/Documents/MTPLX/models/Qwen3.6-27B-MTPLX-Flat8-Quality-TargetBase", - "policy": "mtplx-flat8-quality", - "policy_report": { - "applied_assignments": { - "language_model.lm_head": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.embed_tokens": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.0.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.1.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.10.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.11.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.12.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.13.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.14.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.15.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.16.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.17.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.18.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.19.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.2.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.20.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.21.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.22.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.23.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.24.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.25.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.26.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.27.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.28.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.29.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.3.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.30.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.31.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.32.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.33.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.34.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.35.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.36.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.37.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.38.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.39.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.4.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.40.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.41.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.42.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.43.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.44.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.45.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.46.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.47.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.48.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.49.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.5.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.50.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.51.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.52.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.53.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.54.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.55.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.56.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.57.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.58.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.59.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.6.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.60.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.61.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.62.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.63.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.self_attn.k_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.self_attn.o_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.self_attn.q_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.7.self_attn.v_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.8.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.linear_attn.in_proj_a": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.linear_attn.in_proj_b": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.linear_attn.in_proj_qkv": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.linear_attn.in_proj_z": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.linear_attn.out_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.mlp.down_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.mlp.gate_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "language_model.model.layers.9.mlp.up_proj": { - "bits": 8, - "group_size": 64, - "mode": "affine" - } - }, - "audit": { - "language_model.lm_head": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.lm_head", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.embed_tokens": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.embed_tokens", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.0.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.1.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.10.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.11.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.12.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.13.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.14.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.15.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.16.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.17.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.18.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.19.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.2.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.20.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.21.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.22.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.23.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.24.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.25.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.26.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.27.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.28.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.29.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.3.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.30.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.31.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.32.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.33.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.34.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.35.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.36.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.37.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.38.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.39.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.4.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.40.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.41.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.42.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.43.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.44.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.45.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.46.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.47.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.48.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.49.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.5.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.50.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.51.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.52.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.53.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.54.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.55.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.56.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.57.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.58.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.59.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.6.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.60.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.61.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.62.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.63.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.self_attn.k_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.self_attn.o_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.self_attn.q_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.7.self_attn.v_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.8.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.linear_attn.in_proj_a": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.linear_attn.in_proj_b": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.linear_attn.in_proj_qkv": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.linear_attn.in_proj_z": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.linear_attn.out_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.mlp.down_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.mlp.gate_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - "language_model.model.layers.9.mlp.up_proj": { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - } - }, - "expected_assignments": [ - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.lm_head", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.embed_tokens", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.0.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.1.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.10.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.11.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.12.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.13.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.14.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.15.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.16.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.17.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.18.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.19.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.2.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.20.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.21.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.22.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.23.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.24.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.25.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.26.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.27.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.28.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.29.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.3.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.30.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.31.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.32.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.33.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.34.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.35.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.36.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.37.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.38.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.39.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.4.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.40.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.41.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.42.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.43.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.44.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.45.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.46.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.47.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.48.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.49.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.5.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.50.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.51.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.52.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.53.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.54.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.55.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.56.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.57.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.58.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.59.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.6.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.60.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.61.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.62.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.63.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.k_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.o_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.q_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.7.self_attn.v_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.8.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_a", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_b", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_qkv", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.in_proj_z", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.linear_attn.out_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.mlp.down_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.mlp.gate_proj", - "reason": "flat 8-bit quality text trunk" - }, - { - "action": "quantize", - "group": "text_all_8bit", - "params": { - "bits": 8, - "group_size": 64, - "mode": "affine" - }, - "path": "language_model.model.layers.9.mlp.up_proj", - "reason": "flat 8-bit quality text trunk" - } - ], - "policy": "mtplx-flat8-quality", - "summary": { - "actions": { - "quantize": 498 - }, - "bits": { - "8": 498 - }, - "groups": { - "text_all_8bit": 498 - } - } + "created_at": "2026-06-06T02:50:21.738527+00:00", + "mode": "vision-compatibility-repair", + "mtp_sidecar": { + "path": "mtp/weights.safetensors", + "sha256": "dff5112ffdeff420d38f9227cc00f40dc302e009e70a3e8dc85c3c443b6364ab", + "size": 488790135, + "source": "previous optimized-quality mtp.safetensors", + "tensor_count": 29 }, - "quantization_histogram": { - "bits": { - "8": 498 - }, - "bits_group_size": { - "8bit_g64": 498 - }, - "module_groups": { - "embed_tokens": 1, - "linear_attn": 240, - "lm_head": 1, - "mlp": 192, - "self_attn": 64 - } + "repo_id": "Youssofal/Qwen3.6-27B-MTPLX-Optimized-Quality", + "root_layout": { + "model_vision_shard_present": false, + "root_mtp_safetensors_present": false, + "root_safetensors": [ + "model-00001-of-00006.safetensors", + "model-00002-of-00006.safetensors", + "model-00003-of-00006.safetensors", + "model-00004-of-00006.safetensors", + "model-00005-of-00006.safetensors", + "model-00006-of-00006.safetensors" + ] }, - "source": "/Users/youssof/Documents/MTPLX/.hf-home/hub/models--Qwen--Qwen3.6-27B/snapshots/6a9e13bd6fc8f0983b9b99948120bc37f49c13e9", - "source_path": "/Users/youssof/Documents/MTPLX/.hf-home/hub/models--Qwen--Qwen3.6-27B/snapshots/6a9e13bd6fc8f0983b9b99948120bc37f49c13e9", - "source_sha": "6a9e13bd6fc8f0983b9b99948120bc37f49c13e9", - "vision_retention": { - "vision_bytes": 921460192, - "vision_shard": "model-vision-00001-of-00001.safetensors", - "vision_tensor_count": 333 + "validation": { + "mlx_vlm_image_passed": true, + "mlx_vlm_text_passed": true, + "mlx_vlm_video_passed": true, + "mtplx_contract_status": "runtime contract remains conservative; exactness_baseline is still candidate-promoted-by-user-decision", + "mtplx_depth": 1, + "mtplx_generation_mode": "mtp", + "mtplx_nested_sidecar_generation_text": "HELLO" } -} \ No newline at end of file +}