423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-11-04 18:02:19 +08:00
Code Issues Packages Projects Releases Wiki Activity

[assets] update wechat (#8385)

Browse Source
This commit is contained in:
Yaowei Zheng 2025-06-16 18:23:22 +08:00 committed by GitHub
parent 8e4ac78607
commit 9a2d1dec62
5 changed files with 21 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@ -5,7 +5,7 @@
[![GitHub contributors](https://img.shields.io/github/contributors/hiyouga/LLaMA-Factory?color=orange)](https://github.com/hiyouga/LLaMA-Factory/graphs/contributors)
[![GitHub workflow](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml/badge.svg)](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml)
[![PyPI](https://img.shields.io/pypi/v/llamafactory)](https://pypi.org/project/llamafactory/)
[![Citation](https://img.shields.io/badge/citation-561-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
[![Citation](https://img.shields.io/badge/citation-614-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
[![Docker Pulls](https://img.shields.io/docker/pulls/hiyouga/llamafactory)](https://hub.docker.com/r/hiyouga/llamafactory/tags)
[![Twitter](https://img.shields.io/twitter/follow/llamafactory_ai)](https://twitter.com/llamafactory_ai)
@ -55,7 +55,7 @@ Choose your path:
- **Colab (free)**: https://colab.research.google.com/drive/1eRTPn37ltBbYsISy9Aw2NuI2Aq5CQrD9?usp=sharing
- **Local machine**: Please refer to [usage](#getting-started)
- **PAI-DSW (free trial)**: https://gallery.pai-ml.com/#/preview/deepLearning/nlp/llama_factory
- **Alaya NeW (cloud GPU deal)**: https://docs.alayanew.com/docs/documents/newActivities/llamafactory/?utm_source=LLaMA-Factory
- **Alaya NeW (cloud GPU deal)**: https://docs.alayanew.com/docs/documents/useGuide/LLaMAFactory/mutiple/?utm_source=LLaMA-Factory
> [!NOTE]
> Except for the above links, all other websites are unauthorized third-party websites. Please carefully use them.

6
README_zh.md
View File

@ -5,7 +5,7 @@
[![GitHub contributors](https://img.shields.io/github/contributors/hiyouga/LLaMA-Factory?color=orange)](https://github.com/hiyouga/LLaMA-Factory/graphs/contributors)
[![GitHub workflow](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml/badge.svg)](https://github.com/hiyouga/LLaMA-Factory/actions/workflows/tests.yml)
[![PyPI](https://img.shields.io/pypi/v/llamafactory)](https://pypi.org/project/llamafactory/)
[![Citation](https://img.shields.io/badge/citation-561-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
[![Citation](https://img.shields.io/badge/citation-614-green)](https://scholar.google.com/scholar?cites=12620864006390196564)
[![Docker Pulls](https://img.shields.io/docker/pulls/hiyouga/llamafactory)](https://hub.docker.com/r/hiyouga/llamafactory/tags)
[![Twitter](https://img.shields.io/twitter/follow/llamafactory_ai)](https://twitter.com/llamafactory_ai)
@ -41,7 +41,7 @@
</div>
👋 加入我们的[微信群](assets/wechat.jpg)、[NPU 用户群](assets/wechat_npu.jpg)或 [Alaya NeW 算力优惠群](assets/wechat_alaya.png)。
👋 加入我们的[微信群](assets/wechat.jpg)、[NPU 用户群](assets/wechat_npu.jpg)或 [九章智算云算力优惠群](assets/wechat_alaya.png)。
\[ [English](README.md) | 中文 \]
@ -57,7 +57,7 @@ https://github.com/user-attachments/assets/43b700c6-a178-41db-b1f8-8190a5d3fcfc
- **Colab（免费）**：https://colab.research.google.com/drive/1d5KQtbemerlSDSxZIfAaWXhKr30QypiK?usp=sharing
- **本地机器**：请见[如何使用](#如何使用)
- **PAI-DSW（免费试用）**：https://gallery.pai-ml.com/#/preview/deepLearning/nlp/llama_factory
- **Alaya NeW（算力优惠活动）**：https://docs.alayanew.com/docs/documents/newActivities/llamafactory/?utm_source=LLaMA-Factory
- **九章智算云（算力优惠活动）**：https://docs.alayanew.com/docs/documents/useGuide/LLaMAFactory/mutiple/?utm_source=LLaMA-Factory
> [!NOTE]
> 除上述链接以外的其他网站均为未经许可的第三方网站，请小心甄别。

7
src/llamafactory/hparams/finetuning_args.py
View File

@ -204,7 +204,12 @@ class RLHFArguments:
)
ld_alpha: Optional[float] = field(
default=None,
metadata={"help": "α parameter from the LD-DPO paper, which controls the weighting of the verbose token log-probabilities in responses"},
metadata={
"help": (
"Alpha parameter from the LD-DPO paper, which controls the weighting of"
" the verbose token log-probabilities in responses."
)
},
)

10
src/llamafactory/train/dpo/trainer.py
View File

@ -188,8 +188,9 @@ class CustomDPOTrainer(DPOTrainer):
batch = nested_detach(batch, clone=True) # avoid error
all_logits: torch.Tensor = model(**batch, return_dict=True, use_cache=False).logits.to(torch.float32)
all_logps, valid_length = get_batch_logps(logits=all_logits, labels=batch["labels"],
ld_alpha=(self.ld_alpha if not is_ref_model else None))
all_logps, valid_length = get_batch_logps(
logits=all_logits, labels=batch["labels"], ld_alpha=(self.ld_alpha if not is_ref_model else None)
)
if self.loss_type in ["ipo", "orpo", "simpo"]:
all_logps = all_logps / valid_length
@ -219,8 +220,9 @@ class CustomDPOTrainer(DPOTrainer):
ref_context = nullcontext()
with torch.no_grad(), ref_context:
reference_chosen_logps, reference_rejected_logps, *_ = self.concatenated_forward(ref_model, batch,
is_ref_model=True)
reference_chosen_logps, reference_rejected_logps, *_ = self.concatenated_forward(
ref_model, batch, is_ref_model=True
)
return reference_chosen_logps, reference_rejected_logps

5
src/llamafactory/train/trainer_utils.py
View File

@ -585,7 +585,10 @@ def create_custom_scheduler(
def get_batch_logps(
logits: "torch.Tensor", labels: "torch.Tensor", label_pad_token_id: int = IGNORE_INDEX, ld_alpha: Optional[float] = None
logits: "torch.Tensor",
labels: "torch.Tensor",
label_pad_token_id: int = IGNORE_INDEX,
ld_alpha: Optional[float] = None,
) -> tuple["torch.Tensor", "torch.Tensor"]:
r"""Compute the log probabilities of the given labels under the given logits.