Muse-Glimmer-30B-INT8-AutoRound / quantization_config.json
Minachist's picture
Restore BF16 q_proj so q/k/v share one quant config
3b16b0e verified
Raw
History Blame Contribute Delete
13.3 kB
{
"bits": 8,
"data_type": "int",
"group_size": -1,
"sym": true,
"gradient_accumulate_steps": 2,
"iters": 250,
"low_gpu_mem_usage": true,
"autoround_version": "0.15.0",
"block_name_to_quantize": "model.language_model.layers",
"quant_method": "auto-round",
"packing_format": "auto_round:auto_gptq",
"extra_config": {
"model.language_model.layers.0.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.1.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.2.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.3.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.4.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.5.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.6.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.7.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.8.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.9.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.10.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.11.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.12.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.13.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.14.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.15.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.16.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.17.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.18.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.19.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.20.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.21.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.22.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.23.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.24.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.25.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.26.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.27.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.28.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.29.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.30.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.31.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.32.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.33.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.34.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.35.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.36.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.37.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.38.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.39.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.40.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.41.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.42.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.43.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.44.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.45.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.46.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.47.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.48.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.49.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.50.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.51.self_attn.k_proj": {
"bits": 16
},
"model.language_model.layers.0.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.1.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.2.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.3.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.4.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.5.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.6.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.7.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.8.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.9.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.10.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.11.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.12.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.13.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.14.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.15.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.16.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.17.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.18.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.19.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.20.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.21.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.22.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.23.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.24.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.25.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.26.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.27.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.28.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.29.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.30.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.31.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.32.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.33.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.34.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.35.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.36.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.37.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.38.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.39.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.40.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.41.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.42.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.43.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.44.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.45.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.46.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.47.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.48.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.49.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.50.self_attn.v_proj": {
"bits": 16
},
"model.language_model.layers.51.self_attn.v_proj": {
"bits": 16
},
".*embed_tokens.*": {
"bits": 16
},
".*vision_tower.*": {
"bits": 16
},
".*vision_adapter.*": {
"bits": 16
},
".*vision_projection.*": {
"bits": 16
},
".*self_attn\\.k_proj.*": {
"bits": 16
},
".*self_attn\\.v_proj.*": {
"bits": 16
},
".*self_attn\\.q_proj.*": {
"bits": 16
},
"model.language_model.layers.0.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.1.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.2.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.3.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.4.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.5.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.6.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.7.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.8.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.9.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.10.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.11.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.12.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.13.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.14.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.15.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.16.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.17.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.18.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.19.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.20.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.21.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.22.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.23.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.24.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.25.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.26.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.27.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.28.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.29.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.30.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.31.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.32.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.33.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.34.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.35.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.36.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.37.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.38.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.39.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.40.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.41.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.42.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.43.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.44.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.45.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.46.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.47.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.48.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.49.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.50.self_attn.q_proj": {
"bits": 16
},
"model.language_model.layers.51.self_attn.q_proj": {
"bits": 16
}
}
}