| { | |
| "base_model_name_or_path": "huggyllama/llama-30b", | |
| "bias": "none", | |
| "fan_in_fan_out": false, | |
| "inference_mode": true, | |
| "init_lazy_lora_weights": true, | |
| "is_r_by_svd": true, | |
| "is_r_reuse": true, | |
| "lazy_lora_alpha": 16.0, | |
| "lazy_lora_dropout": 0.05, | |
| "lazy_pre_adapter_type": "none", | |
| "lazy_pre_lora_alpha": 0.1, | |
| "modules_to_save": null, | |
| "num_attention_heads": 52, | |
| "num_layers": 60, | |
| "num_transformer_submodules": 1, | |
| "num_virtual_tokens": null, | |
| "peft_type": "LAZY_LORA", | |
| "prefix_tuning_config": null, | |
| "prompt_tuning_config": null, | |
| "r": 64, | |
| "r_by_module_dict": { | |
| "model.layers.0.mlp.down_proj": 55, | |
| "model.layers.0.mlp.gate_proj": 28, | |
| "model.layers.0.mlp.up_proj": 29, | |
| "model.layers.0.self_attn.k_proj": 36, | |
| "model.layers.0.self_attn.o_proj": 29, | |
| "model.layers.0.self_attn.q_proj": 39, | |
| "model.layers.0.self_attn.v_proj": 28, | |
| "model.layers.1.mlp.down_proj": 61, | |
| "model.layers.1.mlp.gate_proj": 49, | |
| "model.layers.1.mlp.up_proj": 50, | |
| "model.layers.1.self_attn.k_proj": 60, | |
| "model.layers.1.self_attn.o_proj": 36, | |
| "model.layers.1.self_attn.q_proj": 59, | |
| "model.layers.1.self_attn.v_proj": 33, | |
| "model.layers.10.mlp.down_proj": 63, | |
| "model.layers.10.mlp.gate_proj": 64, | |
| "model.layers.10.mlp.up_proj": 64, | |
| "model.layers.10.self_attn.k_proj": 79, | |
| "model.layers.10.self_attn.o_proj": 58, | |
| "model.layers.10.self_attn.q_proj": 78, | |
| "model.layers.10.self_attn.v_proj": 58, | |
| "model.layers.11.mlp.down_proj": 63, | |
| "model.layers.11.mlp.gate_proj": 64, | |
| "model.layers.11.mlp.up_proj": 65, | |
| "model.layers.11.self_attn.k_proj": 77, | |
| "model.layers.11.self_attn.o_proj": 59, | |
| "model.layers.11.self_attn.q_proj": 75, | |
| "model.layers.11.self_attn.v_proj": 59, | |
| "model.layers.12.mlp.down_proj": 62, | |
| "model.layers.12.mlp.gate_proj": 64, | |
| "model.layers.12.mlp.up_proj": 65, | |
| "model.layers.12.self_attn.k_proj": 77, | |
| "model.layers.12.self_attn.o_proj": 59, | |
| "model.layers.12.self_attn.q_proj": 76, | |
| "model.layers.12.self_attn.v_proj": 59, | |
| "model.layers.13.mlp.down_proj": 63, | |
| "model.layers.13.mlp.gate_proj": 63, | |
| "model.layers.13.mlp.up_proj": 65, | |
| "model.layers.13.self_attn.k_proj": 76, | |
| "model.layers.13.self_attn.o_proj": 58, | |
| "model.layers.13.self_attn.q_proj": 75, | |
| "model.layers.13.self_attn.v_proj": 58, | |
| "model.layers.14.mlp.down_proj": 63, | |
| "model.layers.14.mlp.gate_proj": 63, | |
| "model.layers.14.mlp.up_proj": 65, | |
| "model.layers.14.self_attn.k_proj": 76, | |
| "model.layers.14.self_attn.o_proj": 58, | |
| "model.layers.14.self_attn.q_proj": 74, | |
| "model.layers.14.self_attn.v_proj": 59, | |
| "model.layers.15.mlp.down_proj": 63, | |
| "model.layers.15.mlp.gate_proj": 63, | |
| "model.layers.15.mlp.up_proj": 65, | |
| "model.layers.15.self_attn.k_proj": 77, | |
| "model.layers.15.self_attn.o_proj": 61, | |
| "model.layers.15.self_attn.q_proj": 76, | |
| "model.layers.15.self_attn.v_proj": 61, | |
| "model.layers.16.mlp.down_proj": 63, | |
| "model.layers.16.mlp.gate_proj": 63, | |
| "model.layers.16.mlp.up_proj": 65, | |
| "model.layers.16.self_attn.k_proj": 77, | |
| "model.layers.16.self_attn.o_proj": 61, | |
| "model.layers.16.self_attn.q_proj": 76, | |
| "model.layers.16.self_attn.v_proj": 62, | |
| "model.layers.17.mlp.down_proj": 63, | |
| "model.layers.17.mlp.gate_proj": 63, | |
| "model.layers.17.mlp.up_proj": 66, | |
| "model.layers.17.self_attn.k_proj": 75, | |
| "model.layers.17.self_attn.o_proj": 61, | |
| "model.layers.17.self_attn.q_proj": 74, | |
| "model.layers.17.self_attn.v_proj": 62, | |
| "model.layers.18.mlp.down_proj": 64, | |
| "model.layers.18.mlp.gate_proj": 63, | |
| "model.layers.18.mlp.up_proj": 66, | |
| "model.layers.18.self_attn.k_proj": 76, | |
| "model.layers.18.self_attn.o_proj": 61, | |
| "model.layers.18.self_attn.q_proj": 74, | |
| "model.layers.18.self_attn.v_proj": 62, | |
| "model.layers.19.mlp.down_proj": 64, | |
| "model.layers.19.mlp.gate_proj": 63, | |
| "model.layers.19.mlp.up_proj": 66, | |
| "model.layers.19.self_attn.k_proj": 74, | |
| "model.layers.19.self_attn.o_proj": 63, | |
| "model.layers.19.self_attn.q_proj": 73, | |
| "model.layers.19.self_attn.v_proj": 64, | |
| "model.layers.2.mlp.down_proj": 62, | |
| "model.layers.2.mlp.gate_proj": 58, | |
| "model.layers.2.mlp.up_proj": 58, | |
| "model.layers.2.self_attn.k_proj": 75, | |
| "model.layers.2.self_attn.o_proj": 48, | |
| "model.layers.2.self_attn.q_proj": 75, | |
| "model.layers.2.self_attn.v_proj": 44, | |
| "model.layers.20.mlp.down_proj": 64, | |
| "model.layers.20.mlp.gate_proj": 63, | |
| "model.layers.20.mlp.up_proj": 66, | |
| "model.layers.20.self_attn.k_proj": 70, | |
| "model.layers.20.self_attn.o_proj": 63, | |
| "model.layers.20.self_attn.q_proj": 70, | |
| "model.layers.20.self_attn.v_proj": 64, | |
| "model.layers.21.mlp.down_proj": 64, | |
| "model.layers.21.mlp.gate_proj": 63, | |
| "model.layers.21.mlp.up_proj": 66, | |
| "model.layers.21.self_attn.k_proj": 72, | |
| "model.layers.21.self_attn.o_proj": 64, | |
| "model.layers.21.self_attn.q_proj": 71, | |
| "model.layers.21.self_attn.v_proj": 64, | |
| "model.layers.22.mlp.down_proj": 65, | |
| "model.layers.22.mlp.gate_proj": 63, | |
| "model.layers.22.mlp.up_proj": 65, | |
| "model.layers.22.self_attn.k_proj": 67, | |
| "model.layers.22.self_attn.o_proj": 66, | |
| "model.layers.22.self_attn.q_proj": 66, | |
| "model.layers.22.self_attn.v_proj": 66, | |
| "model.layers.23.mlp.down_proj": 65, | |
| "model.layers.23.mlp.gate_proj": 64, | |
| "model.layers.23.mlp.up_proj": 66, | |
| "model.layers.23.self_attn.k_proj": 68, | |
| "model.layers.23.self_attn.o_proj": 65, | |
| "model.layers.23.self_attn.q_proj": 67, | |
| "model.layers.23.self_attn.v_proj": 65, | |
| "model.layers.24.mlp.down_proj": 65, | |
| "model.layers.24.mlp.gate_proj": 64, | |
| "model.layers.24.mlp.up_proj": 65, | |
| "model.layers.24.self_attn.k_proj": 69, | |
| "model.layers.24.self_attn.o_proj": 66, | |
| "model.layers.24.self_attn.q_proj": 69, | |
| "model.layers.24.self_attn.v_proj": 67, | |
| "model.layers.25.mlp.down_proj": 65, | |
| "model.layers.25.mlp.gate_proj": 64, | |
| "model.layers.25.mlp.up_proj": 65, | |
| "model.layers.25.self_attn.k_proj": 72, | |
| "model.layers.25.self_attn.o_proj": 66, | |
| "model.layers.25.self_attn.q_proj": 71, | |
| "model.layers.25.self_attn.v_proj": 66, | |
| "model.layers.26.mlp.down_proj": 65, | |
| "model.layers.26.mlp.gate_proj": 64, | |
| "model.layers.26.mlp.up_proj": 65, | |
| "model.layers.26.self_attn.k_proj": 67, | |
| "model.layers.26.self_attn.o_proj": 68, | |
| "model.layers.26.self_attn.q_proj": 67, | |
| "model.layers.26.self_attn.v_proj": 67, | |
| "model.layers.27.mlp.down_proj": 65, | |
| "model.layers.27.mlp.gate_proj": 64, | |
| "model.layers.27.mlp.up_proj": 64, | |
| "model.layers.27.self_attn.k_proj": 62, | |
| "model.layers.27.self_attn.o_proj": 67, | |
| "model.layers.27.self_attn.q_proj": 62, | |
| "model.layers.27.self_attn.v_proj": 67, | |
| "model.layers.28.mlp.down_proj": 65, | |
| "model.layers.28.mlp.gate_proj": 65, | |
| "model.layers.28.mlp.up_proj": 65, | |
| "model.layers.28.self_attn.k_proj": 64, | |
| "model.layers.28.self_attn.o_proj": 69, | |
| "model.layers.28.self_attn.q_proj": 64, | |
| "model.layers.28.self_attn.v_proj": 69, | |
| "model.layers.29.mlp.down_proj": 65, | |
| "model.layers.29.mlp.gate_proj": 65, | |
| "model.layers.29.mlp.up_proj": 64, | |
| "model.layers.29.self_attn.k_proj": 60, | |
| "model.layers.29.self_attn.o_proj": 69, | |
| "model.layers.29.self_attn.q_proj": 61, | |
| "model.layers.29.self_attn.v_proj": 69, | |
| "model.layers.3.mlp.down_proj": 62, | |
| "model.layers.3.mlp.gate_proj": 61, | |
| "model.layers.3.mlp.up_proj": 60, | |
| "model.layers.3.self_attn.k_proj": 80, | |
| "model.layers.3.self_attn.o_proj": 48, | |
| "model.layers.3.self_attn.q_proj": 78, | |
| "model.layers.3.self_attn.v_proj": 46, | |
| "model.layers.30.mlp.down_proj": 65, | |
| "model.layers.30.mlp.gate_proj": 65, | |
| "model.layers.30.mlp.up_proj": 64, | |
| "model.layers.30.self_attn.k_proj": 62, | |
| "model.layers.30.self_attn.o_proj": 69, | |
| "model.layers.30.self_attn.q_proj": 62, | |
| "model.layers.30.self_attn.v_proj": 69, | |
| "model.layers.31.mlp.down_proj": 65, | |
| "model.layers.31.mlp.gate_proj": 65, | |
| "model.layers.31.mlp.up_proj": 64, | |
| "model.layers.31.self_attn.k_proj": 60, | |
| "model.layers.31.self_attn.o_proj": 69, | |
| "model.layers.31.self_attn.q_proj": 61, | |
| "model.layers.31.self_attn.v_proj": 68, | |
| "model.layers.32.mlp.down_proj": 65, | |
| "model.layers.32.mlp.gate_proj": 66, | |
| "model.layers.32.mlp.up_proj": 64, | |
| "model.layers.32.self_attn.k_proj": 59, | |
| "model.layers.32.self_attn.o_proj": 69, | |
| "model.layers.32.self_attn.q_proj": 60, | |
| "model.layers.32.self_attn.v_proj": 69, | |
| "model.layers.33.mlp.down_proj": 65, | |
| "model.layers.33.mlp.gate_proj": 66, | |
| "model.layers.33.mlp.up_proj": 64, | |
| "model.layers.33.self_attn.k_proj": 65, | |
| "model.layers.33.self_attn.o_proj": 70, | |
| "model.layers.33.self_attn.q_proj": 65, | |
| "model.layers.33.self_attn.v_proj": 69, | |
| "model.layers.34.mlp.down_proj": 65, | |
| "model.layers.34.mlp.gate_proj": 66, | |
| "model.layers.34.mlp.up_proj": 64, | |
| "model.layers.34.self_attn.k_proj": 63, | |
| "model.layers.34.self_attn.o_proj": 69, | |
| "model.layers.34.self_attn.q_proj": 63, | |
| "model.layers.34.self_attn.v_proj": 69, | |
| "model.layers.35.mlp.down_proj": 65, | |
| "model.layers.35.mlp.gate_proj": 66, | |
| "model.layers.35.mlp.up_proj": 64, | |
| "model.layers.35.self_attn.k_proj": 61, | |
| "model.layers.35.self_attn.o_proj": 69, | |
| "model.layers.35.self_attn.q_proj": 61, | |
| "model.layers.35.self_attn.v_proj": 69, | |
| "model.layers.36.mlp.down_proj": 65, | |
| "model.layers.36.mlp.gate_proj": 66, | |
| "model.layers.36.mlp.up_proj": 64, | |
| "model.layers.36.self_attn.k_proj": 63, | |
| "model.layers.36.self_attn.o_proj": 70, | |
| "model.layers.36.self_attn.q_proj": 63, | |
| "model.layers.36.self_attn.v_proj": 70, | |
| "model.layers.37.mlp.down_proj": 65, | |
| "model.layers.37.mlp.gate_proj": 66, | |
| "model.layers.37.mlp.up_proj": 64, | |
| "model.layers.37.self_attn.k_proj": 59, | |
| "model.layers.37.self_attn.o_proj": 70, | |
| "model.layers.37.self_attn.q_proj": 60, | |
| "model.layers.37.self_attn.v_proj": 70, | |
| "model.layers.38.mlp.down_proj": 65, | |
| "model.layers.38.mlp.gate_proj": 66, | |
| "model.layers.38.mlp.up_proj": 64, | |
| "model.layers.38.self_attn.k_proj": 57, | |
| "model.layers.38.self_attn.o_proj": 71, | |
| "model.layers.38.self_attn.q_proj": 58, | |
| "model.layers.38.self_attn.v_proj": 71, | |
| "model.layers.39.mlp.down_proj": 65, | |
| "model.layers.39.mlp.gate_proj": 66, | |
| "model.layers.39.mlp.up_proj": 64, | |
| "model.layers.39.self_attn.k_proj": 57, | |
| "model.layers.39.self_attn.o_proj": 70, | |
| "model.layers.39.self_attn.q_proj": 58, | |
| "model.layers.39.self_attn.v_proj": 70, | |
| "model.layers.4.mlp.down_proj": 62, | |
| "model.layers.4.mlp.gate_proj": 63, | |
| "model.layers.4.mlp.up_proj": 62, | |
| "model.layers.4.self_attn.k_proj": 77, | |
| "model.layers.4.self_attn.o_proj": 53, | |
| "model.layers.4.self_attn.q_proj": 76, | |
| "model.layers.4.self_attn.v_proj": 51, | |
| "model.layers.40.mlp.down_proj": 65, | |
| "model.layers.40.mlp.gate_proj": 67, | |
| "model.layers.40.mlp.up_proj": 65, | |
| "model.layers.40.self_attn.k_proj": 57, | |
| "model.layers.40.self_attn.o_proj": 72, | |
| "model.layers.40.self_attn.q_proj": 57, | |
| "model.layers.40.self_attn.v_proj": 71, | |
| "model.layers.41.mlp.down_proj": 65, | |
| "model.layers.41.mlp.gate_proj": 66, | |
| "model.layers.41.mlp.up_proj": 65, | |
| "model.layers.41.self_attn.k_proj": 54, | |
| "model.layers.41.self_attn.o_proj": 71, | |
| "model.layers.41.self_attn.q_proj": 55, | |
| "model.layers.41.self_attn.v_proj": 71, | |
| "model.layers.42.mlp.down_proj": 65, | |
| "model.layers.42.mlp.gate_proj": 66, | |
| "model.layers.42.mlp.up_proj": 65, | |
| "model.layers.42.self_attn.k_proj": 52, | |
| "model.layers.42.self_attn.o_proj": 71, | |
| "model.layers.42.self_attn.q_proj": 53, | |
| "model.layers.42.self_attn.v_proj": 71, | |
| "model.layers.43.mlp.down_proj": 65, | |
| "model.layers.43.mlp.gate_proj": 67, | |
| "model.layers.43.mlp.up_proj": 65, | |
| "model.layers.43.self_attn.k_proj": 58, | |
| "model.layers.43.self_attn.o_proj": 71, | |
| "model.layers.43.self_attn.q_proj": 58, | |
| "model.layers.43.self_attn.v_proj": 71, | |
| "model.layers.44.mlp.down_proj": 65, | |
| "model.layers.44.mlp.gate_proj": 67, | |
| "model.layers.44.mlp.up_proj": 65, | |
| "model.layers.44.self_attn.k_proj": 55, | |
| "model.layers.44.self_attn.o_proj": 71, | |
| "model.layers.44.self_attn.q_proj": 56, | |
| "model.layers.44.self_attn.v_proj": 71, | |
| "model.layers.45.mlp.down_proj": 65, | |
| "model.layers.45.mlp.gate_proj": 67, | |
| "model.layers.45.mlp.up_proj": 65, | |
| "model.layers.45.self_attn.k_proj": 55, | |
| "model.layers.45.self_attn.o_proj": 71, | |
| "model.layers.45.self_attn.q_proj": 56, | |
| "model.layers.45.self_attn.v_proj": 71, | |
| "model.layers.46.mlp.down_proj": 65, | |
| "model.layers.46.mlp.gate_proj": 67, | |
| "model.layers.46.mlp.up_proj": 65, | |
| "model.layers.46.self_attn.k_proj": 50, | |
| "model.layers.46.self_attn.o_proj": 68, | |
| "model.layers.46.self_attn.q_proj": 52, | |
| "model.layers.46.self_attn.v_proj": 69, | |
| "model.layers.47.mlp.down_proj": 65, | |
| "model.layers.47.mlp.gate_proj": 67, | |
| "model.layers.47.mlp.up_proj": 65, | |
| "model.layers.47.self_attn.k_proj": 50, | |
| "model.layers.47.self_attn.o_proj": 70, | |
| "model.layers.47.self_attn.q_proj": 52, | |
| "model.layers.47.self_attn.v_proj": 71, | |
| "model.layers.48.mlp.down_proj": 65, | |
| "model.layers.48.mlp.gate_proj": 67, | |
| "model.layers.48.mlp.up_proj": 65, | |
| "model.layers.48.self_attn.k_proj": 53, | |
| "model.layers.48.self_attn.o_proj": 71, | |
| "model.layers.48.self_attn.q_proj": 55, | |
| "model.layers.48.self_attn.v_proj": 71, | |
| "model.layers.49.mlp.down_proj": 65, | |
| "model.layers.49.mlp.gate_proj": 67, | |
| "model.layers.49.mlp.up_proj": 66, | |
| "model.layers.49.self_attn.k_proj": 57, | |
| "model.layers.49.self_attn.o_proj": 73, | |
| "model.layers.49.self_attn.q_proj": 58, | |
| "model.layers.49.self_attn.v_proj": 74, | |
| "model.layers.5.mlp.down_proj": 62, | |
| "model.layers.5.mlp.gate_proj": 64, | |
| "model.layers.5.mlp.up_proj": 63, | |
| "model.layers.5.self_attn.k_proj": 76, | |
| "model.layers.5.self_attn.o_proj": 53, | |
| "model.layers.5.self_attn.q_proj": 75, | |
| "model.layers.5.self_attn.v_proj": 52, | |
| "model.layers.50.mlp.down_proj": 65, | |
| "model.layers.50.mlp.gate_proj": 67, | |
| "model.layers.50.mlp.up_proj": 66, | |
| "model.layers.50.self_attn.k_proj": 56, | |
| "model.layers.50.self_attn.o_proj": 72, | |
| "model.layers.50.self_attn.q_proj": 57, | |
| "model.layers.50.self_attn.v_proj": 72, | |
| "model.layers.51.mlp.down_proj": 65, | |
| "model.layers.51.mlp.gate_proj": 67, | |
| "model.layers.51.mlp.up_proj": 66, | |
| "model.layers.51.self_attn.k_proj": 57, | |
| "model.layers.51.self_attn.o_proj": 70, | |
| "model.layers.51.self_attn.q_proj": 58, | |
| "model.layers.51.self_attn.v_proj": 71, | |
| "model.layers.52.mlp.down_proj": 65, | |
| "model.layers.52.mlp.gate_proj": 66, | |
| "model.layers.52.mlp.up_proj": 66, | |
| "model.layers.52.self_attn.k_proj": 54, | |
| "model.layers.52.self_attn.o_proj": 70, | |
| "model.layers.52.self_attn.q_proj": 55, | |
| "model.layers.52.self_attn.v_proj": 70, | |
| "model.layers.53.mlp.down_proj": 65, | |
| "model.layers.53.mlp.gate_proj": 66, | |
| "model.layers.53.mlp.up_proj": 66, | |
| "model.layers.53.self_attn.k_proj": 54, | |
| "model.layers.53.self_attn.o_proj": 68, | |
| "model.layers.53.self_attn.q_proj": 56, | |
| "model.layers.53.self_attn.v_proj": 69, | |
| "model.layers.54.mlp.down_proj": 66, | |
| "model.layers.54.mlp.gate_proj": 66, | |
| "model.layers.54.mlp.up_proj": 67, | |
| "model.layers.54.self_attn.k_proj": 55, | |
| "model.layers.54.self_attn.o_proj": 70, | |
| "model.layers.54.self_attn.q_proj": 56, | |
| "model.layers.54.self_attn.v_proj": 71, | |
| "model.layers.55.mlp.down_proj": 66, | |
| "model.layers.55.mlp.gate_proj": 66, | |
| "model.layers.55.mlp.up_proj": 67, | |
| "model.layers.55.self_attn.k_proj": 56, | |
| "model.layers.55.self_attn.o_proj": 70, | |
| "model.layers.55.self_attn.q_proj": 57, | |
| "model.layers.55.self_attn.v_proj": 70, | |
| "model.layers.56.mlp.down_proj": 65, | |
| "model.layers.56.mlp.gate_proj": 66, | |
| "model.layers.56.mlp.up_proj": 67, | |
| "model.layers.56.self_attn.k_proj": 53, | |
| "model.layers.56.self_attn.o_proj": 73, | |
| "model.layers.56.self_attn.q_proj": 54, | |
| "model.layers.56.self_attn.v_proj": 74, | |
| "model.layers.57.mlp.down_proj": 66, | |
| "model.layers.57.mlp.gate_proj": 66, | |
| "model.layers.57.mlp.up_proj": 67, | |
| "model.layers.57.self_attn.k_proj": 54, | |
| "model.layers.57.self_attn.o_proj": 68, | |
| "model.layers.57.self_attn.q_proj": 55, | |
| "model.layers.57.self_attn.v_proj": 69, | |
| "model.layers.58.mlp.down_proj": 65, | |
| "model.layers.58.mlp.gate_proj": 67, | |
| "model.layers.58.mlp.up_proj": 67, | |
| "model.layers.58.self_attn.k_proj": 49, | |
| "model.layers.58.self_attn.o_proj": 63, | |
| "model.layers.58.self_attn.q_proj": 50, | |
| "model.layers.58.self_attn.v_proj": 65, | |
| "model.layers.59.mlp.down_proj": 65, | |
| "model.layers.59.mlp.gate_proj": 68, | |
| "model.layers.59.mlp.up_proj": 68, | |
| "model.layers.59.self_attn.k_proj": 53, | |
| "model.layers.59.self_attn.o_proj": 57, | |
| "model.layers.59.self_attn.q_proj": 53, | |
| "model.layers.59.self_attn.v_proj": 60, | |
| "model.layers.6.mlp.down_proj": 62, | |
| "model.layers.6.mlp.gate_proj": 64, | |
| "model.layers.6.mlp.up_proj": 63, | |
| "model.layers.6.self_attn.k_proj": 76, | |
| "model.layers.6.self_attn.o_proj": 54, | |
| "model.layers.6.self_attn.q_proj": 75, | |
| "model.layers.6.self_attn.v_proj": 53, | |
| "model.layers.7.mlp.down_proj": 62, | |
| "model.layers.7.mlp.gate_proj": 64, | |
| "model.layers.7.mlp.up_proj": 64, | |
| "model.layers.7.self_attn.k_proj": 78, | |
| "model.layers.7.self_attn.o_proj": 56, | |
| "model.layers.7.self_attn.q_proj": 77, | |
| "model.layers.7.self_attn.v_proj": 55, | |
| "model.layers.8.mlp.down_proj": 62, | |
| "model.layers.8.mlp.gate_proj": 64, | |
| "model.layers.8.mlp.up_proj": 64, | |
| "model.layers.8.self_attn.k_proj": 80, | |
| "model.layers.8.self_attn.o_proj": 58, | |
| "model.layers.8.self_attn.q_proj": 78, | |
| "model.layers.8.self_attn.v_proj": 58, | |
| "model.layers.9.mlp.down_proj": 62, | |
| "model.layers.9.mlp.gate_proj": 64, | |
| "model.layers.9.mlp.up_proj": 64, | |
| "model.layers.9.self_attn.k_proj": 80, | |
| "model.layers.9.self_attn.o_proj": 58, | |
| "model.layers.9.self_attn.q_proj": 78, | |
| "model.layers.9.self_attn.v_proj": 58 | |
| }, | |
| "target_modules": [ | |
| "down_proj", | |
| "gate_proj", | |
| "k_proj", | |
| "o_proj", | |
| "v_proj", | |
| "q_proj", | |
| "up_proj" | |
| ], | |
| "task_type": "CAUSAL_LM", | |
| "token_dim": 6656 | |
| } |