423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2026-03-10 13:56:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix lora target

Browse Source 
Former-commit-id: d822e41e7ac7e310ee49e347fc45754284ce30b8
This commit is contained in:
hiyouga
2023-09-09 17:04:45 +08:00
parent 34da72ffbb
commit 9a3b4d9ee0
7 changed files with 63 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llmtuner/tuner/core/adapter.py
View File

@@ -82,7 +82,7 @@ def init_adapter(
model = PeftModel.from_pretrained(model, latest_checkpoint, is_trainable=is_trainable)
if is_trainable and latest_checkpoint is None: # create new lora weights while training
if len(finetuning_args.lora_target) == 1 and finetuning_args.lora_target == "all":
if len(finetuning_args.lora_target) == 1 and finetuning_args.lora_target[0] == "all":
target_modules = find_all_linear_modules(model, model_args.quantization_bit)
else:
target_modules = finetuning_args.lora_target

2
src/llmtuner/tuner/core/utils.py
View File

@@ -20,7 +20,7 @@ def find_all_linear_modules(
module_names = set()
for name, module in model.named_modules():
if isinstance(module, linear_cls):
if output_layer_name not in name and isinstance(module, linear_cls):
module_names.add(name.split(".")[-1])
if output_layer_name in module_names: