Skip to content

Commit

Permalink
Merge remote-tracking branch 'paddlenlp/develop' into dev_fix_qwen2_i…
Browse files Browse the repository at this point in the history
…mport
  • Loading branch information
DrownFish19 committed Jul 10, 2024
2 parents 2e7b7cb + fbe613b commit d779dd7
Show file tree
Hide file tree
Showing 27 changed files with 930 additions and 942 deletions.
3 changes: 1 addition & 2 deletions docs/llm/docs/peft.md
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,6 @@ Prefix-tuning[论文](https://arxiv.org/abs/2101.00190)
target_modules=target_modules,
r=lora_rank,
lora_alpha=2 * lora_rank,
merge_weights=True
)
model = LoRAModel(model, lora_config)
model.mark_only_lora_as_trainable()
Expand Down Expand Up @@ -92,7 +91,7 @@ Parameters:
默认为 0.0,dropout的比例设置,float 类型
--merge_weights
默认为 False,模型推理时,是否进行base model 权重和 LoRA 权重的合参操作,bool 类型
默认为 False,接口将被废弃。请使用model.merge()或model.unmerge()替代。
--trainable_bias
指定可训练的 bias, 可选项 ['lora', 'all']
Expand Down
Loading

0 comments on commit d779dd7

Please sign in to comment.