|
{ |
|
"alpha_pattern": {}, |
|
"auto_mapping": { |
|
"base_model_class": "GPTModel", |
|
"parent_library": "models.gpt" |
|
}, |
|
"base_model_name_or_path": "adammoss/gpt-pretrain-lm-w1", |
|
"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": 8, |
|
"lora_dropout": 0.1, |
|
"megatron_config": null, |
|
"megatron_core": "megatron.core", |
|
"modules_to_save": [ |
|
"class_head" |
|
], |
|
"peft_type": "LORA", |
|
"r": 4, |
|
"rank_pattern": {}, |
|
"revision": null, |
|
"target_modules": [ |
|
"blocks.2.sa.heads.2.value", |
|
"blocks.4.sa.heads.3.query", |
|
"blocks.0.sa.heads.1.value", |
|
"blocks.5.sa.heads.4.query", |
|
"blocks.1.sa.heads.0.query", |
|
"blocks.4.sa.heads.4.value", |
|
"blocks.0.sa.heads.0.query", |
|
"blocks.0.sa.heads.4.query", |
|
"blocks.0.sa.heads.0.value", |
|
"blocks.2.sa.heads.1.query", |
|
"blocks.0.sa.heads.5.value", |
|
"blocks.5.sa.heads.0.value", |
|
"blocks.4.sa.heads.2.query", |
|
"blocks.4.sa.heads.1.value", |
|
"blocks.2.sa.heads.5.query", |
|
"blocks.2.sa.heads.2.query", |
|
"blocks.3.sa.heads.2.query", |
|
"blocks.2.sa.heads.1.value", |
|
"blocks.4.sa.heads.4.query", |
|
"blocks.5.sa.heads.5.value", |
|
"blocks.3.sa.heads.0.value", |
|
"blocks.0.sa.heads.2.query", |
|
"blocks.1.sa.heads.5.value", |
|
"blocks.5.sa.heads.2.value", |
|
"blocks.0.sa.heads.1.query", |
|
"blocks.0.sa.heads.3.query", |
|
"blocks.2.sa.heads.3.value", |
|
"blocks.3.sa.heads.1.value", |
|
"blocks.5.sa.heads.3.value", |
|
"blocks.2.sa.heads.5.value", |
|
"blocks.5.sa.heads.1.value", |
|
"blocks.3.sa.heads.4.query", |
|
"blocks.2.sa.heads.0.query", |
|
"blocks.1.sa.heads.1.query", |
|
"blocks.4.sa.heads.0.value", |
|
"blocks.1.sa.heads.1.value", |
|
"blocks.0.sa.heads.3.value", |
|
"blocks.5.sa.heads.4.value", |
|
"blocks.5.sa.heads.5.query", |
|
"blocks.1.sa.heads.5.query", |
|
"blocks.3.sa.heads.0.query", |
|
"blocks.3.sa.heads.5.value", |
|
"blocks.0.sa.heads.5.query", |
|
"blocks.3.sa.heads.2.value", |
|
"blocks.0.sa.heads.4.value", |
|
"blocks.5.sa.heads.1.query", |
|
"blocks.2.sa.heads.0.value", |
|
"blocks.0.sa.heads.2.value", |
|
"blocks.4.sa.heads.0.query", |
|
"blocks.3.sa.heads.3.value", |
|
"blocks.2.sa.heads.3.query", |
|
"blocks.4.sa.heads.5.query", |
|
"blocks.4.sa.heads.2.value", |
|
"blocks.1.sa.heads.2.query", |
|
"blocks.5.sa.heads.0.query", |
|
"blocks.1.sa.heads.4.query", |
|
"blocks.2.sa.heads.4.value", |
|
"blocks.5.sa.heads.2.query", |
|
"blocks.1.sa.heads.2.value", |
|
"blocks.3.sa.heads.5.query", |
|
"blocks.2.sa.heads.4.query", |
|
"blocks.1.sa.heads.3.value", |
|
"blocks.4.sa.heads.3.value", |
|
"blocks.3.sa.heads.1.query", |
|
"blocks.1.sa.heads.3.query", |
|
"blocks.3.sa.heads.3.query", |
|
"blocks.3.sa.heads.4.value", |
|
"blocks.4.sa.heads.1.query", |
|
"blocks.1.sa.heads.0.value", |
|
"blocks.4.sa.heads.5.value", |
|
"blocks.5.sa.heads.3.query", |
|
"blocks.1.sa.heads.4.value" |
|
], |
|
"task_type": null, |
|
"use_dora": false, |
|
"use_rslora": false |
|
} |