{ "_name_or_path": "/data1/DeepSeek-R1-bf16", "architectures": [ "DeepseekV3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_deepseek.DeepseekV3Config", "AutoModel": "modeling_deepseek.DeepseekV3Model", "AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM" }, "aux_loss_alpha": 0.001, "bos_token_id": 0, "eos_token_id": 1, "ep_size": 1, "first_k_dense_replace": 3, "hidden_act": "silu", "hidden_size": 7168, "initializer_range": 0.02, "intermediate_size": 18432, "kv_lora_rank": 512, "max_position_embeddings": 163840, "model_type": "deepseek_v3", "moe_intermediate_size": 2048, "moe_layer_freq": 1, "n_group": 8, "n_routed_experts": 256, "n_shared_experts": 1, "norm_topk_prob": true, "num_attention_heads": 128, "num_experts_per_tok": 8, "num_hidden_layers": 61, "num_key_value_heads": 128, "num_nextn_predict_layers": 1, "pretraining_tp": 1, "q_lora_rank": 1536, "qk_nope_head_dim": 128, "qk_rope_head_dim": 64, "quantization_config": { "amp": true, "autoround_version": "0.4.5", "backend": "auto_round:gptq:exllamav2", "batch_size": 4, "bits": 2, "data_type": "int", "dataset": "NeelNanda/pile-10k", "enable_minmax_tuning": true, "enable_norm_bias_tuning": false, "enable_quanted_input": true, "extra_config": { "model.layers.0.mlp.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.0.mlp.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.0.mlp.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.0.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.0.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.0.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.0.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.0.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.mlp.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.mlp.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.mlp.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.1.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.1.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.10.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.10.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.11.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.11.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.12.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.12.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.13.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.13.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.14.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.14.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.15.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.15.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.16.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.16.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.17.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.17.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.18.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.18.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.19.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.19.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.mlp.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.mlp.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.mlp.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.2.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.2.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.20.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.20.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.21.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.21.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.22.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.22.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.23.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.23.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.24.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.24.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.25.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.25.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.26.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.26.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.27.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.27.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.28.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.28.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.29.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.29.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.3.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.3.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.30.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.30.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.31.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.31.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.32.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.32.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.33.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.33.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.34.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.34.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.35.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.35.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.36.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.36.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.37.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.37.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.38.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.38.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.39.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.39.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.4.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.4.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.40.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.40.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.41.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.41.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.42.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.42.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.43.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.43.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.44.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.44.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.45.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.45.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.46.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.46.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.47.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.47.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.48.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.48.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.49.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.49.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.5.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.5.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.50.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.50.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.51.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.51.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.52.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.52.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.53.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.53.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.54.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.54.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.55.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.55.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.56.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.56.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.57.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.57.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.58.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.58.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.59.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.59.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.6.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.6.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.60.mlp.experts.0.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.1.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.10.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.100.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.101.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.102.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.103.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.104.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.105.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.106.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.107.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.108.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.109.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.11.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.110.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.111.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.112.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.113.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.114.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.115.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.116.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.117.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.118.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.119.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.12.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.120.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.121.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.122.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.123.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.124.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.125.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.126.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.127.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.128.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.129.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.13.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.130.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.131.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.132.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.133.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.134.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.135.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.136.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.137.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.138.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.139.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.14.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.140.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.141.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.142.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.143.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.144.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.145.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.146.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.147.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.148.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.149.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.15.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.150.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.151.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.152.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.153.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.154.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.155.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.156.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.157.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.158.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.159.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.16.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.160.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.161.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.162.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.163.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.164.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.165.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.166.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.167.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.168.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.169.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.17.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.170.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.171.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.172.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.173.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.174.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.175.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.176.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.177.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.178.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.179.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.18.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.180.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.181.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.182.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.183.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.184.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.185.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.186.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.187.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.188.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.189.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.19.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.190.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.191.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.192.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.193.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.194.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.195.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.196.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.197.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.198.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.199.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.2.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.20.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.200.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.201.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.202.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.203.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.204.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.205.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.206.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.207.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.208.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.209.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.21.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.210.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.211.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.212.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.213.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.214.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.215.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.216.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.217.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.218.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.219.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.22.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.220.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.221.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.222.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.223.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.224.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.225.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.226.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.227.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.228.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.229.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.23.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.230.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.231.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.232.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.233.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.234.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.235.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.236.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.237.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.238.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.239.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.24.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.240.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.241.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.242.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.243.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.244.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.245.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.246.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.247.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.248.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.249.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.25.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.250.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.251.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.252.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.253.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.254.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.255.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.26.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.27.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.28.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.29.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.3.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.30.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.31.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.32.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.33.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.34.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.35.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.36.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.37.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.38.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.39.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.4.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.40.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.41.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.42.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.43.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.44.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.45.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.46.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.47.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.48.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.49.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.5.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.50.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.51.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.52.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.53.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.54.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.55.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.56.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.57.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.58.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.59.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.6.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.60.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.61.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.62.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.63.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.64.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.65.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.66.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.67.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.68.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.69.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.7.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.70.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.71.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.72.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.73.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.74.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.75.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.76.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.77.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.78.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.79.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.8.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.80.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.81.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.82.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.83.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.84.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.85.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.86.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.87.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.88.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.89.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.9.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.90.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.91.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.92.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.93.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.94.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.95.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.96.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.97.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.98.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.experts.99.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.shared_experts.down_proj": { "bits": 16, "data_type": "bfloat" }, "model.layers.60.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.60.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.60.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.60.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.60.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.60.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.60.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.7.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.7.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.8.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.8.self_attn.q_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.mlp.shared_experts.down_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.mlp.shared_experts.gate_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.mlp.shared_experts.up_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.self_attn.kv_a_proj_with_mqa": { "bits": 4, "group_size": 128 }, "model.layers.9.self_attn.kv_b_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.self_attn.o_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.self_attn.q_a_proj": { "bits": 4, "group_size": 128 }, "model.layers.9.self_attn.q_b_proj": { "bits": 4, "group_size": 128 } }, "gradient_accumulate_steps": 1, "group_size": 64, "iters": 400, "low_gpu_mem_usage": false, "lr": 0.0025, "minmax_lr": 0.0025, "nsamples": 512, "quant_method": "intel/auto-round", "scale_dtype": "torch.float16", "seqlen": 512, "sym": true, "to_quant_block_names": null }, "rms_norm_eps": 1e-06, "rope_scaling": { "beta_fast": 32, "beta_slow": 1, "factor": 40, "mscale": 1.0, "mscale_all_dim": 1.0, "original_max_position_embeddings": 4096, "type": "yarn" }, "rope_theta": 10000, "routed_scaling_factor": 2.5, "scoring_func": "sigmoid", "seq_aux": true, "tie_word_embeddings": false, "topk_group": 4, "topk_method": "noaux_tc", "torch_dtype": "float16", "transformers_version": "4.47.0", "use_cache": true, "v_head_dim": 128, "vocab_size": 129280 }