pretrain_wo-cot_w-asd_gpt / adapter_config.json
slplab's picture
Training in progress, step 1
819a53d verified
raw
history blame
6.57 kB
{
"alpha_pattern": {},
"auto_mapping": null,
"base_model_name_or_path": "kakaobrain/kogpt",
"bias": "none",
"fan_in_fan_out": false,
"inference_mode": true,
"init_lora_weights": true,
"layer_replication": null,
"layers_pattern": null,
"layers_to_transform": null,
"loftq_config": {},
"lora_alpha": 16,
"lora_dropout": 0.05,
"megatron_config": null,
"megatron_core": "megatron.core",
"modules_to_save": null,
"peft_type": "LORA",
"r": 8,
"rank_pattern": {},
"revision": null,
"target_modules": [
"transformer.h.17.mlp.fc_out",
"transformer.h.8.mlp.fc_in",
"transformer.h.11.attn.k_proj",
"transformer.h.17.attn.q_proj",
"transformer.h.1.mlp.fc_in",
"transformer.h.1.mlp.fc_out",
"transformer.h.16.mlp.fc_out",
"transformer.h.5.attn.out_proj",
"transformer.h.15.attn.k_proj",
"transformer.h.0.attn.q_proj",
"transformer.h.5.attn.k_proj",
"transformer.h.14.attn.out_proj",
"transformer.h.10.attn.q_proj",
"transformer.h.26.attn.out_proj",
"transformer.h.6.attn.v_proj",
"transformer.h.8.attn.v_proj",
"transformer.h.7.attn.k_proj",
"transformer.h.22.attn.q_proj",
"transformer.h.13.attn.out_proj",
"transformer.h.7.attn.q_proj",
"transformer.h.26.attn.k_proj",
"transformer.h.1.attn.out_proj",
"transformer.h.15.mlp.fc_out",
"transformer.h.3.attn.out_proj",
"transformer.h.24.mlp.fc_in",
"transformer.h.11.attn.q_proj",
"transformer.h.21.attn.q_proj",
"transformer.h.6.attn.k_proj",
"transformer.h.13.attn.v_proj",
"transformer.h.17.attn.out_proj",
"transformer.h.24.attn.k_proj",
"transformer.h.25.attn.q_proj",
"transformer.h.18.mlp.fc_out",
"transformer.h.11.attn.v_proj",
"transformer.h.8.attn.out_proj",
"transformer.h.2.attn.k_proj",
"transformer.h.10.attn.out_proj",
"transformer.h.15.attn.out_proj",
"transformer.h.27.attn.k_proj",
"transformer.h.5.mlp.fc_out",
"transformer.h.0.mlp.fc_out",
"transformer.h.21.mlp.fc_in",
"transformer.h.3.mlp.fc_out",
"transformer.h.18.attn.k_proj",
"transformer.h.14.mlp.fc_in",
"transformer.h.4.attn.k_proj",
"transformer.h.23.attn.v_proj",
"transformer.h.9.attn.q_proj",
"transformer.h.19.attn.out_proj",
"transformer.h.5.attn.v_proj",
"transformer.h.25.attn.v_proj",
"transformer.h.15.attn.q_proj",
"transformer.h.2.attn.out_proj",
"transformer.h.23.mlp.fc_in",
"transformer.h.3.mlp.fc_in",
"transformer.h.16.attn.v_proj",
"transformer.h.3.attn.v_proj",
"transformer.h.27.mlp.fc_out",
"transformer.h.14.attn.v_proj",
"transformer.h.8.attn.k_proj",
"transformer.h.26.attn.v_proj",
"transformer.h.25.mlp.fc_out",
"transformer.h.5.attn.q_proj",
"transformer.h.16.mlp.fc_in",
"transformer.h.18.mlp.fc_in",
"transformer.h.14.attn.q_proj",
"transformer.h.5.mlp.fc_in",
"transformer.h.8.attn.q_proj",
"transformer.h.9.attn.k_proj",
"transformer.h.27.attn.v_proj",
"transformer.h.18.attn.v_proj",
"transformer.h.24.attn.v_proj",
"transformer.h.0.mlp.fc_in",
"transformer.h.22.attn.v_proj",
"transformer.h.22.mlp.fc_out",
"transformer.h.20.attn.out_proj",
"transformer.h.7.attn.v_proj",
"transformer.h.16.attn.q_proj",
"transformer.h.3.attn.q_proj",
"transformer.h.3.attn.k_proj",
"transformer.h.25.mlp.fc_in",
"transformer.h.21.attn.k_proj",
"transformer.h.9.mlp.fc_out",
"transformer.h.22.attn.out_proj",
"transformer.h.1.attn.q_proj",
"transformer.h.2.mlp.fc_in",
"transformer.h.20.attn.q_proj",
"transformer.h.21.attn.v_proj",
"transformer.h.6.mlp.fc_in",
"transformer.h.8.mlp.fc_out",
"transformer.h.26.attn.q_proj",
"transformer.h.17.mlp.fc_in",
"transformer.h.0.attn.out_proj",
"transformer.h.23.attn.out_proj",
"transformer.h.10.attn.v_proj",
"transformer.h.13.mlp.fc_out",
"transformer.h.25.attn.out_proj",
"transformer.h.7.mlp.fc_in",
"transformer.h.10.mlp.fc_in",
"transformer.h.26.mlp.fc_in",
"transformer.h.21.attn.out_proj",
"transformer.h.13.attn.k_proj",
"transformer.h.14.attn.k_proj",
"transformer.h.11.mlp.fc_out",
"transformer.h.16.attn.out_proj",
"transformer.h.6.attn.q_proj",
"transformer.h.14.mlp.fc_out",
"transformer.h.20.mlp.fc_out",
"transformer.h.22.attn.k_proj",
"transformer.h.20.attn.v_proj",
"transformer.h.19.mlp.fc_out",
"transformer.h.22.mlp.fc_in",
"transformer.h.4.attn.out_proj",
"transformer.h.12.mlp.fc_in",
"transformer.h.19.attn.v_proj",
"transformer.h.15.attn.v_proj",
"transformer.h.7.attn.out_proj",
"transformer.h.4.attn.q_proj",
"transformer.h.12.attn.k_proj",
"transformer.h.24.attn.q_proj",
"transformer.h.2.attn.v_proj",
"transformer.h.11.mlp.fc_in",
"transformer.h.10.mlp.fc_out",
"transformer.h.18.attn.out_proj",
"transformer.h.2.mlp.fc_out",
"transformer.h.20.attn.k_proj",
"transformer.h.26.mlp.fc_out",
"transformer.h.0.attn.k_proj",
"transformer.h.12.attn.v_proj",
"transformer.h.4.attn.v_proj",
"transformer.h.12.mlp.fc_out",
"transformer.h.17.attn.k_proj",
"transformer.h.18.attn.q_proj",
"transformer.h.27.attn.out_proj",
"transformer.h.0.attn.v_proj",
"transformer.h.4.mlp.fc_out",
"transformer.h.2.attn.q_proj",
"transformer.h.23.attn.k_proj",
"transformer.h.25.attn.k_proj",
"transformer.h.1.attn.k_proj",
"transformer.h.13.mlp.fc_in",
"transformer.h.27.attn.q_proj",
"transformer.h.6.mlp.fc_out",
"transformer.h.20.mlp.fc_in",
"transformer.h.9.mlp.fc_in",
"transformer.h.19.mlp.fc_in",
"transformer.h.24.attn.out_proj",
"transformer.h.12.attn.q_proj",
"transformer.h.23.mlp.fc_out",
"transformer.h.6.attn.out_proj",
"transformer.h.13.attn.q_proj",
"transformer.h.12.attn.out_proj",
"transformer.h.19.attn.q_proj",
"transformer.h.15.mlp.fc_in",
"transformer.h.23.attn.q_proj",
"transformer.h.9.attn.out_proj",
"transformer.h.19.attn.k_proj",
"transformer.h.17.attn.v_proj",
"transformer.h.21.mlp.fc_out",
"transformer.h.7.mlp.fc_out",
"transformer.h.11.attn.out_proj",
"transformer.h.10.attn.k_proj",
"transformer.h.1.attn.v_proj",
"transformer.h.4.mlp.fc_in",
"transformer.h.9.attn.v_proj",
"transformer.h.24.mlp.fc_out",
"transformer.h.27.mlp.fc_in",
"transformer.h.16.attn.k_proj"
],
"task_type": "CAUSAL_LM",
"use_dora": false,
"use_rslora": false
}