{
  "architectures": [
    "Glm4MoeLiteForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 0,
  "dtype": "bfloat16",
  "eos_token_id": [
    154820,
    154827,
    154829
  ],
  "first_k_dense_replace": 1,
  "head_dim": 64,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 10240,
  "kv_lora_rank": 512,
  "max_position_embeddings": 202752,
  "mlp_layer_types": [
    "dense",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse",
    "sparse"
  ],
  "model_type": "glm4_moe_lite",
  "moe_intermediate_size": 1536,
  "n_group": 1,
  "n_routed_experts": 48,
  "n_shared_experts": 1,
  "norm_topk_prob": true,
  "num_attention_heads": 20,
  "num_experts_per_tok": 4,
  "num_hidden_layers": 47,
  "num_key_value_heads": 20,
  "num_nextn_predict_layers": 0,
  "pad_token_id": 154820,
  "partial_rotary_factor": 1.0,
  "pretraining_tp": 1,
  "q_lora_rank": 768,
  "qk_head_dim": 256,
  "qk_nope_head_dim": 192,
  "qk_rope_head_dim": 64,
  "rms_norm_eps": 1e-05,
  "rope_interleave": true,
  "rope_parameters": {
    "partial_rotary_factor": 1.0,
    "rope_theta": 1000000,
    "rope_type": "default"
  },
  "routed_scaling_factor": 1.8,
  "tie_word_embeddings": false,
  "topk_group": 1,
  "topk_method": "noaux_tc",
  "transformers_version": "5.0.0rc3",
  "use_cache": true,
  "v_head_dim": 256,
  "vocab_size": 154880,
  "quantization": {
    "group_size": 64,
    "bits": 5,
    "mode": "affine",
    "model.layers.5.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.5.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.0.mlp.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.mlp.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.mlp.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.4.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.embed_tokens": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.9.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.9.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.10.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.10.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.15.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.13.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.14.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.13.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.11.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.37.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.41.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.45.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.44.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.44.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.45.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "lm_head": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.0.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.4.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.5.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.5.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.6.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.10.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.11.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.11.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.13.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.14.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    }
  },
  "quantization_config": {
    "group_size": 64,
    "bits": 5,
    "mode": "affine",
    "model.layers.5.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.5.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.0.mlp.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.mlp.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.mlp.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.4.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.embed_tokens": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.9.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.9.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.10.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.5.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.10.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.15.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.13.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.14.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.13.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.11.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.12.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.11.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.37.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.41.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.38.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.42.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.self_attn.q_b_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.self_attn.o_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.mlp.shared_experts.down_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.45.self_attn.o_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.mlp.shared_experts.gate_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.44.mlp.shared_experts.down_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.mlp.shared_experts.up_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.self_attn.q_a_proj": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.44.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.mlp.shared_experts.up_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.q_b_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.self_attn.kv_a_proj_with_mqa": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.45.mlp.shared_experts.gate_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.q_a_proj": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.kv_a_proj_with_mqa": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "lm_head": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.0.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.0.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.1.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.1.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.2.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.3.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.4.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.4.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.5.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.5.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.6.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.6.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.7.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.8.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.9.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.10.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.10.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.11.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.11.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.12.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.13.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.13.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.14.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.14.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.15.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.16.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.17.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.18.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.19.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.20.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.21.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.22.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.23.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.24.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.25.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.26.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.27.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.28.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.29.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.30.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.31.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.32.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.33.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.34.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.35.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.36.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.37.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.38.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.39.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.39.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.40.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.40.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.41.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.42.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.43.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.44.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.embed_q": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.45.self_attn.unembed_out": {
      "bits": 6,
      "group_size": 64,
      "mode": "affine"
    },
    "model.layers.46.self_attn.embed_q": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    },
    "model.layers.46.self_attn.unembed_out": {
      "bits": 8,
      "group_size": 32,
      "mode": "mxfp8"
    }
  }
}