lazylora-13bhf / adapter_config.json
xianchaowu
lazy lora for llama2 13b hf
4900ca9
raw
history blame
12.6 kB
{
"base_model_name_or_path": "meta-llama/Llama-2-13b-hf",
"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": 40,
"num_layers": 40,
"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": 56,
"model.layers.0.mlp.gate_proj": 45,
"model.layers.0.mlp.up_proj": 47,
"model.layers.0.self_attn.k_proj": 14,
"model.layers.0.self_attn.o_proj": 24,
"model.layers.0.self_attn.q_proj": 13,
"model.layers.0.self_attn.v_proj": 30,
"model.layers.1.mlp.down_proj": 61,
"model.layers.1.mlp.gate_proj": 51,
"model.layers.1.mlp.up_proj": 55,
"model.layers.1.self_attn.k_proj": 39,
"model.layers.1.self_attn.o_proj": 36,
"model.layers.1.self_attn.q_proj": 39,
"model.layers.1.self_attn.v_proj": 38,
"model.layers.10.mlp.down_proj": 62,
"model.layers.10.mlp.gate_proj": 63,
"model.layers.10.mlp.up_proj": 64,
"model.layers.10.self_attn.k_proj": 68,
"model.layers.10.self_attn.o_proj": 56,
"model.layers.10.self_attn.q_proj": 69,
"model.layers.10.self_attn.v_proj": 55,
"model.layers.11.mlp.down_proj": 63,
"model.layers.11.mlp.gate_proj": 62,
"model.layers.11.mlp.up_proj": 64,
"model.layers.11.self_attn.k_proj": 71,
"model.layers.11.self_attn.o_proj": 55,
"model.layers.11.self_attn.q_proj": 70,
"model.layers.11.self_attn.v_proj": 55,
"model.layers.12.mlp.down_proj": 63,
"model.layers.12.mlp.gate_proj": 62,
"model.layers.12.mlp.up_proj": 65,
"model.layers.12.self_attn.k_proj": 68,
"model.layers.12.self_attn.o_proj": 56,
"model.layers.12.self_attn.q_proj": 67,
"model.layers.12.self_attn.v_proj": 57,
"model.layers.13.mlp.down_proj": 64,
"model.layers.13.mlp.gate_proj": 62,
"model.layers.13.mlp.up_proj": 65,
"model.layers.13.self_attn.k_proj": 64,
"model.layers.13.self_attn.o_proj": 58,
"model.layers.13.self_attn.q_proj": 65,
"model.layers.13.self_attn.v_proj": 58,
"model.layers.14.mlp.down_proj": 64,
"model.layers.14.mlp.gate_proj": 62,
"model.layers.14.mlp.up_proj": 65,
"model.layers.14.self_attn.k_proj": 68,
"model.layers.14.self_attn.o_proj": 58,
"model.layers.14.self_attn.q_proj": 68,
"model.layers.14.self_attn.v_proj": 58,
"model.layers.15.mlp.down_proj": 65,
"model.layers.15.mlp.gate_proj": 62,
"model.layers.15.mlp.up_proj": 66,
"model.layers.15.self_attn.k_proj": 67,
"model.layers.15.self_attn.o_proj": 59,
"model.layers.15.self_attn.q_proj": 66,
"model.layers.15.self_attn.v_proj": 60,
"model.layers.16.mlp.down_proj": 65,
"model.layers.16.mlp.gate_proj": 62,
"model.layers.16.mlp.up_proj": 66,
"model.layers.16.self_attn.k_proj": 67,
"model.layers.16.self_attn.o_proj": 60,
"model.layers.16.self_attn.q_proj": 66,
"model.layers.16.self_attn.v_proj": 61,
"model.layers.17.mlp.down_proj": 65,
"model.layers.17.mlp.gate_proj": 63,
"model.layers.17.mlp.up_proj": 65,
"model.layers.17.self_attn.k_proj": 67,
"model.layers.17.self_attn.o_proj": 63,
"model.layers.17.self_attn.q_proj": 67,
"model.layers.17.self_attn.v_proj": 62,
"model.layers.18.mlp.down_proj": 65,
"model.layers.18.mlp.gate_proj": 63,
"model.layers.18.mlp.up_proj": 65,
"model.layers.18.self_attn.k_proj": 69,
"model.layers.18.self_attn.o_proj": 65,
"model.layers.18.self_attn.q_proj": 69,
"model.layers.18.self_attn.v_proj": 65,
"model.layers.19.mlp.down_proj": 65,
"model.layers.19.mlp.gate_proj": 64,
"model.layers.19.mlp.up_proj": 65,
"model.layers.19.self_attn.k_proj": 66,
"model.layers.19.self_attn.o_proj": 65,
"model.layers.19.self_attn.q_proj": 66,
"model.layers.19.self_attn.v_proj": 64,
"model.layers.2.mlp.down_proj": 62,
"model.layers.2.mlp.gate_proj": 59,
"model.layers.2.mlp.up_proj": 60,
"model.layers.2.self_attn.k_proj": 60,
"model.layers.2.self_attn.o_proj": 50,
"model.layers.2.self_attn.q_proj": 60,
"model.layers.2.self_attn.v_proj": 49,
"model.layers.20.mlp.down_proj": 65,
"model.layers.20.mlp.gate_proj": 64,
"model.layers.20.mlp.up_proj": 65,
"model.layers.20.self_attn.k_proj": 67,
"model.layers.20.self_attn.o_proj": 65,
"model.layers.20.self_attn.q_proj": 68,
"model.layers.20.self_attn.v_proj": 64,
"model.layers.21.mlp.down_proj": 65,
"model.layers.21.mlp.gate_proj": 64,
"model.layers.21.mlp.up_proj": 65,
"model.layers.21.self_attn.k_proj": 65,
"model.layers.21.self_attn.o_proj": 67,
"model.layers.21.self_attn.q_proj": 65,
"model.layers.21.self_attn.v_proj": 66,
"model.layers.22.mlp.down_proj": 65,
"model.layers.22.mlp.gate_proj": 65,
"model.layers.22.mlp.up_proj": 65,
"model.layers.22.self_attn.k_proj": 68,
"model.layers.22.self_attn.o_proj": 71,
"model.layers.22.self_attn.q_proj": 68,
"model.layers.22.self_attn.v_proj": 70,
"model.layers.23.mlp.down_proj": 65,
"model.layers.23.mlp.gate_proj": 65,
"model.layers.23.mlp.up_proj": 64,
"model.layers.23.self_attn.k_proj": 65,
"model.layers.23.self_attn.o_proj": 71,
"model.layers.23.self_attn.q_proj": 66,
"model.layers.23.self_attn.v_proj": 69,
"model.layers.24.mlp.down_proj": 65,
"model.layers.24.mlp.gate_proj": 66,
"model.layers.24.mlp.up_proj": 64,
"model.layers.24.self_attn.k_proj": 67,
"model.layers.24.self_attn.o_proj": 71,
"model.layers.24.self_attn.q_proj": 67,
"model.layers.24.self_attn.v_proj": 70,
"model.layers.25.mlp.down_proj": 65,
"model.layers.25.mlp.gate_proj": 66,
"model.layers.25.mlp.up_proj": 64,
"model.layers.25.self_attn.k_proj": 66,
"model.layers.25.self_attn.o_proj": 73,
"model.layers.25.self_attn.q_proj": 67,
"model.layers.25.self_attn.v_proj": 72,
"model.layers.26.mlp.down_proj": 65,
"model.layers.26.mlp.gate_proj": 66,
"model.layers.26.mlp.up_proj": 64,
"model.layers.26.self_attn.k_proj": 67,
"model.layers.26.self_attn.o_proj": 73,
"model.layers.26.self_attn.q_proj": 67,
"model.layers.26.self_attn.v_proj": 73,
"model.layers.27.mlp.down_proj": 65,
"model.layers.27.mlp.gate_proj": 66,
"model.layers.27.mlp.up_proj": 65,
"model.layers.27.self_attn.k_proj": 68,
"model.layers.27.self_attn.o_proj": 73,
"model.layers.27.self_attn.q_proj": 68,
"model.layers.27.self_attn.v_proj": 72,
"model.layers.28.mlp.down_proj": 65,
"model.layers.28.mlp.gate_proj": 67,
"model.layers.28.mlp.up_proj": 65,
"model.layers.28.self_attn.k_proj": 67,
"model.layers.28.self_attn.o_proj": 73,
"model.layers.28.self_attn.q_proj": 68,
"model.layers.28.self_attn.v_proj": 73,
"model.layers.29.mlp.down_proj": 65,
"model.layers.29.mlp.gate_proj": 67,
"model.layers.29.mlp.up_proj": 65,
"model.layers.29.self_attn.k_proj": 69,
"model.layers.29.self_attn.o_proj": 74,
"model.layers.29.self_attn.q_proj": 69,
"model.layers.29.self_attn.v_proj": 73,
"model.layers.3.mlp.down_proj": 62,
"model.layers.3.mlp.gate_proj": 63,
"model.layers.3.mlp.up_proj": 62,
"model.layers.3.self_attn.k_proj": 66,
"model.layers.3.self_attn.o_proj": 52,
"model.layers.3.self_attn.q_proj": 64,
"model.layers.3.self_attn.v_proj": 53,
"model.layers.30.mlp.down_proj": 65,
"model.layers.30.mlp.gate_proj": 67,
"model.layers.30.mlp.up_proj": 65,
"model.layers.30.self_attn.k_proj": 68,
"model.layers.30.self_attn.o_proj": 77,
"model.layers.30.self_attn.q_proj": 69,
"model.layers.30.self_attn.v_proj": 76,
"model.layers.31.mlp.down_proj": 65,
"model.layers.31.mlp.gate_proj": 67,
"model.layers.31.mlp.up_proj": 65,
"model.layers.31.self_attn.k_proj": 69,
"model.layers.31.self_attn.o_proj": 74,
"model.layers.31.self_attn.q_proj": 69,
"model.layers.31.self_attn.v_proj": 74,
"model.layers.32.mlp.down_proj": 65,
"model.layers.32.mlp.gate_proj": 67,
"model.layers.32.mlp.up_proj": 66,
"model.layers.32.self_attn.k_proj": 66,
"model.layers.32.self_attn.o_proj": 77,
"model.layers.32.self_attn.q_proj": 67,
"model.layers.32.self_attn.v_proj": 76,
"model.layers.33.mlp.down_proj": 65,
"model.layers.33.mlp.gate_proj": 67,
"model.layers.33.mlp.up_proj": 66,
"model.layers.33.self_attn.k_proj": 66,
"model.layers.33.self_attn.o_proj": 75,
"model.layers.33.self_attn.q_proj": 67,
"model.layers.33.self_attn.v_proj": 75,
"model.layers.34.mlp.down_proj": 65,
"model.layers.34.mlp.gate_proj": 67,
"model.layers.34.mlp.up_proj": 67,
"model.layers.34.self_attn.k_proj": 64,
"model.layers.34.self_attn.o_proj": 77,
"model.layers.34.self_attn.q_proj": 65,
"model.layers.34.self_attn.v_proj": 77,
"model.layers.35.mlp.down_proj": 66,
"model.layers.35.mlp.gate_proj": 67,
"model.layers.35.mlp.up_proj": 67,
"model.layers.35.self_attn.k_proj": 64,
"model.layers.35.self_attn.o_proj": 77,
"model.layers.35.self_attn.q_proj": 64,
"model.layers.35.self_attn.v_proj": 76,
"model.layers.36.mlp.down_proj": 67,
"model.layers.36.mlp.gate_proj": 67,
"model.layers.36.mlp.up_proj": 67,
"model.layers.36.self_attn.k_proj": 63,
"model.layers.36.self_attn.o_proj": 78,
"model.layers.36.self_attn.q_proj": 63,
"model.layers.36.self_attn.v_proj": 77,
"model.layers.37.mlp.down_proj": 67,
"model.layers.37.mlp.gate_proj": 67,
"model.layers.37.mlp.up_proj": 67,
"model.layers.37.self_attn.k_proj": 58,
"model.layers.37.self_attn.o_proj": 79,
"model.layers.37.self_attn.q_proj": 59,
"model.layers.37.self_attn.v_proj": 78,
"model.layers.38.mlp.down_proj": 66,
"model.layers.38.mlp.gate_proj": 68,
"model.layers.38.mlp.up_proj": 67,
"model.layers.38.self_attn.k_proj": 57,
"model.layers.38.self_attn.o_proj": 81,
"model.layers.38.self_attn.q_proj": 57,
"model.layers.38.self_attn.v_proj": 81,
"model.layers.39.mlp.down_proj": 66,
"model.layers.39.mlp.gate_proj": 71,
"model.layers.39.mlp.up_proj": 68,
"model.layers.39.self_attn.k_proj": 55,
"model.layers.39.self_attn.o_proj": 75,
"model.layers.39.self_attn.q_proj": 55,
"model.layers.39.self_attn.v_proj": 74,
"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": 66,
"model.layers.4.self_attn.o_proj": 52,
"model.layers.4.self_attn.q_proj": 66,
"model.layers.4.self_attn.v_proj": 52,
"model.layers.5.mlp.down_proj": 62,
"model.layers.5.mlp.gate_proj": 64,
"model.layers.5.mlp.up_proj": 62,
"model.layers.5.self_attn.k_proj": 64,
"model.layers.5.self_attn.o_proj": 53,
"model.layers.5.self_attn.q_proj": 64,
"model.layers.5.self_attn.v_proj": 53,
"model.layers.6.mlp.down_proj": 61,
"model.layers.6.mlp.gate_proj": 65,
"model.layers.6.mlp.up_proj": 62,
"model.layers.6.self_attn.k_proj": 69,
"model.layers.6.self_attn.o_proj": 55,
"model.layers.6.self_attn.q_proj": 68,
"model.layers.6.self_attn.v_proj": 55,
"model.layers.7.mlp.down_proj": 61,
"model.layers.7.mlp.gate_proj": 65,
"model.layers.7.mlp.up_proj": 62,
"model.layers.7.self_attn.k_proj": 69,
"model.layers.7.self_attn.o_proj": 55,
"model.layers.7.self_attn.q_proj": 69,
"model.layers.7.self_attn.v_proj": 56,
"model.layers.8.mlp.down_proj": 62,
"model.layers.8.mlp.gate_proj": 64,
"model.layers.8.mlp.up_proj": 63,
"model.layers.8.self_attn.k_proj": 68,
"model.layers.8.self_attn.o_proj": 56,
"model.layers.8.self_attn.q_proj": 67,
"model.layers.8.self_attn.v_proj": 56,
"model.layers.9.mlp.down_proj": 62,
"model.layers.9.mlp.gate_proj": 64,
"model.layers.9.mlp.up_proj": 63,
"model.layers.9.self_attn.k_proj": 69,
"model.layers.9.self_attn.o_proj": 56,
"model.layers.9.self_attn.q_proj": 69,
"model.layers.9.self_attn.v_proj": 56
},
"rank_file": "",
"target_modules": [
"gate_proj",
"k_proj",
"q_proj",
"down_proj",
"up_proj",
"v_proj",
"o_proj"
],
"task_type": "CAUSAL_LM",
"token_dim": 5120
}