| { |
| "base_model": "microsoft/phi-4", |
| "format": "dashq-packed-linear", |
| "format_version": 1, |
| "method": "dashq", |
| "model_class": "causal_lm", |
| "params": { |
| "bits": 3, |
| "group_size": 128, |
| "low_memory_optimization": false, |
| "moe_hessian_scope": "shared", |
| "n_samples": 128, |
| "scale_zero_dtype": "float16", |
| "symmetric": false, |
| "use_error_compensation": true, |
| "use_optimal_shrinkage": true, |
| "use_weighted_quantization": true |
| }, |
| "quantized_modules": { |
| "model.layers.0.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.0.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.0.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.0.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.1.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.1.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.1.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.1.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.10.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.10.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.10.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.10.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.11.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.11.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.11.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.11.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.12.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.12.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.12.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.12.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.13.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.13.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.13.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.13.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.14.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.14.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.14.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.14.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.15.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.15.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.15.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.15.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.16.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.16.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.16.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.16.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.17.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.17.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.17.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.17.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.18.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.18.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.18.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.18.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.19.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.19.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.19.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.19.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.2.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.2.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.2.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.2.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.20.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.20.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.20.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.20.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.21.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.21.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.21.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.21.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.22.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.22.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.22.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.22.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.23.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.23.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.23.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.23.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.24.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.24.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.24.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.24.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.25.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.25.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.25.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.25.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.26.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.26.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.26.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.26.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.27.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.27.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.27.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.27.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.28.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.28.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.28.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.28.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.29.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.29.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.29.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.29.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.3.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.3.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.3.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.3.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.30.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.30.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.30.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.30.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.31.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.31.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.31.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.31.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.32.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.32.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.32.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.32.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.33.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.33.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.33.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.33.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.34.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.34.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.34.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.34.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.35.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.35.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.35.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.35.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.36.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.36.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.36.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.36.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.37.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.37.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.37.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.37.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.38.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.38.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.38.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.38.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.39.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.39.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.39.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.39.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.4.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.4.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.4.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.4.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.5.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.5.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.5.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.5.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.6.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.6.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.6.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.6.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.7.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.7.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.7.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.7.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.8.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.8.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.8.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.8.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.9.mlp.down_proj": { |
| "group_size": 128, |
| "in_features": 17920, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 716800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 17920, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.9.mlp.gate_up_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 1433600, |
| "out_features": 35840, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.9.self_attn.o_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 204800, |
| "out_features": 5120, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| }, |
| "model.layers.9.self_attn.qkv_proj": { |
| "group_size": 128, |
| "in_features": 5120, |
| "linear_dtype": "bfloat16", |
| "nbits": 3, |
| "num_groups": 307200, |
| "out_features": 7680, |
| "packing": "int3_packed_u32", |
| "quant_in_features": 5120, |
| "runtime_backend": "torch", |
| "scale_zero_dtype": "float16" |
| } |
| }, |
| "results": { |
| "Method": "dashq", |
| "Model": "microsoft/phi-4", |
| "ModelSizeGB": 7.934682704, |
| "OriginalSizeGB": 29.319042992, |
| "PPL": 6.784809589385986, |
| "Params": "{'bits': 3, 'group_size': 128, 'scale_zero_dtype': 'float16', 'n_samples': 128, 'moe_hessian_scope': 'shared', 'use_error_compensation': True, 'use_optimal_shrinkage': True, 'use_weighted_quantization': True, 'symmetric': False, 'low_memory_optimization': False}", |
| "QuantTime": 992.3389053344727, |
| "arc_challenge": 55.631399317406135, |
| "arc_easy": 74.87373737373737, |
| "commonsense_qa": 76.4946764946765, |
| "hellaswag": 80.12348137821151, |
| "lambada_openai": 74.26741703861828, |
| "openbookqa": 45.0, |
| "piqa": 81.39281828073993, |
| "truthfulqa_mc2": 57.791403450384216, |
| "winogrande": 75.13812154696133, |
| "zeroshot_avg": 68.96811720897058 |
| } |
| } |
