423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-19 12:12:48 +08:00
Code Issues Packages Projects Releases Wiki Activity

[assets] update wechat (#8962)

Browse Source
This commit is contained in:
Yaowei Zheng 2025-08-19 02:55:09 +08:00 committed by GitHub
parent 842595698b
commit 44beb76cdf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
10 changed files with 29 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
README.md
View File

@ -10,7 +10,6 @@
[![Twitter](https://img.shields.io/twitter/follow/llamafactory_ai)](https://twitter.com/llamafactory_ai)
[![Discord](https://dcbadge.vercel.app/api/server/rKfvV9r9FK?compact=true&style=flat)](https://discord.gg/rKfvV9r9FK)
[![GitCode](https://gitcode.com/zhengyaowei/LLaMA-Factory/star/badge.svg)](https://gitcode.com/zhengyaowei/LLaMA-Factory)
[![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1eRTPn37ltBbYsISy9Aw2NuI2Aq5CQrD9?usp=sharing)
[![Open in DSW](https://gallery.pai-ml.com/assets/open-in-dsw.svg)](https://gallery.pai-ml.com/#/preview/deepLearning/nlp/llama_factory)
@ -101,7 +100,7 @@ Choose your path:
## Blogs
- [Fine-tune GPT-OSS for Role-Playing using LLaMA-Factory](https://docs.llamafactory.com.cn/docs/documents/best-practice/gptoss/?utm_source=LLaMA-Factory) (Chinese)
- [Fine-tune GPT-OSS for Role-Playing using LLaMA-Factory](https://docs.llamafactory.com.cn/docs/documents/best-practice/gptroleplay/?utm_source=LLaMA-Factory) (Chinese)
- [Fine-tune Llama3.1-70B for Medical Diagnosis using LLaMA-Factory](https://docs.alayanew.com/docs/documents/bestPractice/bigModel/llama70B/?utm_source=LLaMA-Factory) (Chinese)
- [A One-Stop Code-Free Model Reinforcement Learning and Deployment Platform based on LLaMA-Factory and EasyR1](https://aws.amazon.com/cn/blogs/china/building-llm-model-hub-based-on-llamafactory-and-easyr1/) (Chinese)
- [How Apoidea Group enhances visual information extraction from banking documents with multimodal models using LLaMA-Factory on Amazon SageMaker HyperPod](https://aws.amazon.com/cn/blogs/machine-learning/how-apoidea-group-enhances-visual-information-extraction-from-banking-documents-with-multimodal-models-using-llama-factory-on-amazon-sagemaker-hyperpod/) (English)
@ -266,7 +265,7 @@ Choose your path:
| [Falcon](https://huggingface.co/tiiuae) | 7B/11B/40B/180B | falcon |
| [Falcon-H1](https://huggingface.co/tiiuae) | 0.5B/1.5B/3B/7B/34B | falcon_h1 |
| [Gemma/Gemma 2/CodeGemma](https://huggingface.co/google) | 2B/7B/9B/27B | gemma/gemma2 |
| [Gemma 3/Gemma 3n](https://huggingface.co/google) | 1B/4B/6B/8B/12B/27B | gemma3/gemma3n |
| [Gemma 3/Gemma 3n](https://huggingface.co/google) | 270M/1B/4B/6B/8B/12B/27B | gemma3/gemma3n |
| [GLM-4/GLM-4-0414/GLM-Z1](https://huggingface.co/zai-org) | 9B/32B | glm4/glmz1 |
| [GLM-4.1V](https://huggingface.co/zai-org) | 9B | glm4v |
| [GLM-4.5/GLM-4.5V](https://huggingface.co/zai-org)* | 106B/355B | glm4_moe/glm4v_moe |

5
README_zh.md
View File

@ -10,7 +10,6 @@
[![Twitter](https://img.shields.io/twitter/follow/llamafactory_ai)](https://twitter.com/llamafactory_ai)
[![Discord](https://dcbadge.vercel.app/api/server/rKfvV9r9FK?compact=true&style=flat)](https://discord.gg/rKfvV9r9FK)
[![GitCode](https://gitcode.com/zhengyaowei/LLaMA-Factory/star/badge.svg)](https://gitcode.com/zhengyaowei/LLaMA-Factory)
[![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1d5KQtbemerlSDSxZIfAaWXhKr30QypiK?usp=sharing)
[![Open in DSW](https://gallery.pai-ml.com/assets/open-in-dsw.svg)](https://gallery.pai-ml.com/#/preview/deepLearning/nlp/llama_factory)
@ -103,7 +102,7 @@ https://github.com/user-attachments/assets/43b700c6-a178-41db-b1f8-8190a5d3fcfc
## 官方博客
- [使用 LLaMA-Factory 构建 GPT-OSS 角色扮演模型](https://docs.llamafactory.com.cn/docs/documents/best-practice/gptoss/?utm_source=LLaMA-Factory)（中文）
- [使用 LLaMA-Factory 构建 GPT-OSS 角色扮演模型](https://docs.llamafactory.com.cn/docs/documents/best-practice/gptroleplay/?utm_source=LLaMA-Factory)（中文）
- [使用 LLaMA-Factory 微调 Llama3.1-70B 医学诊断模型](https://docs.alayanew.com/docs/documents/bestPractice/bigModel/llama70B/?utm_source=LLaMA-Factory)（中文）
- [基于 LLaMA-Factory 和 EasyR1 打造一站式无代码大模型强化学习和部署平台 LLM Model Hub](https://aws.amazon.com/cn/blogs/china/building-llm-model-hub-based-on-llamafactory-and-easyr1/)（中文）
- [通过亚马逊 SageMaker HyperPod 上的 LLaMA-Factory 增强多模态模型银行文档的视觉信息提取](https://aws.amazon.com/cn/blogs/machine-learning/how-apoidea-group-enhances-visual-information-extraction-from-banking-documents-with-multimodal-models-using-llama-factory-on-amazon-sagemaker-hyperpod/)（英文）
@ -268,7 +267,7 @@ https://github.com/user-attachments/assets/43b700c6-a178-41db-b1f8-8190a5d3fcfc
| [Falcon](https://huggingface.co/tiiuae) | 7B/11B/40B/180B | falcon |
| [Falcon-H1](https://huggingface.co/tiiuae) | 0.5B/1.5B/3B/7B/34B | falcon_h1 |
| [Gemma/Gemma 2/CodeGemma](https://huggingface.co/google) | 2B/7B/9B/27B | gemma/gemma2 |
| [Gemma 3/Gemma 3n](https://huggingface.co/google) | 1B/4B/6B/8B/12B/27B | gemma3/gemma3n |
| [Gemma 3/Gemma 3n](https://huggingface.co/google) | 270M/1B/4B/6B/8B/12B/27B | gemma3/gemma3n |
| [GLM-4/GLM-4-0414/GLM-Z1](https://huggingface.co/zai-org) | 9B/32B | glm4/glmz1 |
| [GLM-4.1V](https://huggingface.co/zai-org) | 9B | glm4v |
| [GLM-4.5/GLM-4.5V](https://huggingface.co/zai-org)* | 106B/355B | glm4_moe/glm4v_moe |

BIN
assets/wechat.jpg
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 169 KiB

After

Width:  |  Height:  |  Size: 167 KiB

BIN
assets/wechat_npu.jpg
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 170 KiB

After

Width:  |  Height:  |  Size: 169 KiB

2
pyproject.toml
View File

@ -4,11 +4,11 @@ build-backend = "setuptools.build_meta"
[project]
name = "llamafactory"
requires-python = ">=3.9.0"
dynamic = [
"version",
"dependencies",
"optional-dependencies",
"requires-python",
"scripts",
"authors",
"description",

2
requirements.txt
View File

@ -6,7 +6,7 @@ peft>=0.14.0,<=0.15.2
trl>=0.8.6,<=0.9.6
tokenizers>=0.19.0,<=0.21.1
# gui
gradio>=4.38.0,<=5.31.0
gradio>=4.38.0,<=5.42.0
matplotlib>=3.7.0
tyro<0.9.0
# ops

14
src/llamafactory/extras/constants.py
View File

@ -776,6 +776,10 @@ register_model_group(
register_model_group(
models={
"Gemma-3-270M": {
DownloadSource.DEFAULT: "google/gemma-3-270m",
DownloadSource.MODELSCOPE: "LLM-Research/gemma-3-270m",
},
"Gemma-3-4B": {
DownloadSource.DEFAULT: "google/gemma-3-4b-pt",
DownloadSource.MODELSCOPE: "LLM-Research/gemma-3-4b-pt",
@ -788,6 +792,10 @@ register_model_group(
DownloadSource.DEFAULT: "google/gemma-3-27b-pt",
DownloadSource.MODELSCOPE: "LLM-Research/gemma-3-27b-pt",
},
"Gemma-3-270M-Instruct": {
DownloadSource.DEFAULT: "google/gemma-3-270m-it",
DownloadSource.MODELSCOPE: "LLM-Research/gemma-3-270m-it",
},
"Gemma-3-4B-Instruct": {
DownloadSource.DEFAULT: "google/gemma-3-4b-it",
DownloadSource.MODELSCOPE: "LLM-Research/gemma-3-4b-it",
@ -1669,8 +1677,8 @@ register_model_group(
},
"MiMo-VL-7B-RL-2508": {
DownloadSource.DEFAULT: "XiaomiMiMo/MiMo-VL-7B-RL-2508",
DownloadSource.MODELSCOPE: "XiaomiMiMo/MiMo-VL-7B-RL-2508"
}
DownloadSource.MODELSCOPE: "XiaomiMiMo/MiMo-VL-7B-RL-2508",
},
},
template="mimo_vl",
multimodal=True,
@ -1685,7 +1693,7 @@ register_model_group(
},
"MiMo-VL-7B-SFT-2508": {
DownloadSource.DEFAULT: "XiaomiMiMo/MiMo-VL-7B-SFT-2508",
DownloadSource.DEFAULT: "XiaomiMiMo/MiMo-VL-7B-SFT-2508"
DownloadSource.DEFAULT: "XiaomiMiMo/MiMo-VL-7B-SFT-2508",
},
},
template="qwen2_vl",

4
src/llamafactory/hparams/parser.py
View File

@ -32,6 +32,7 @@ from transformers.utils import is_torch_bf16_gpu_available, is_torch_npu_availab
from ..extras import logging
from ..extras.constants import CHECKPOINT_NAMES, EngineName
from ..extras.misc import check_dependencies, check_version, get_current_device, is_env_enabled
from ..extras.packages import is_transformers_version_greater_than
from .data_args import DataArguments
from .evaluation_args import EvaluationArguments
from .finetuning_args import FinetuningArguments
@ -304,6 +305,9 @@ def get_train_args(args: Optional[Union[dict[str, Any], list[str]]] = None) -> _
if model_args.use_unsloth and is_deepspeed_zero3_enabled():
raise ValueError("Unsloth is incompatible with DeepSpeed ZeRO-3.")
if data_args.neat_packing and is_transformers_version_greater_than("4.53.0"):
raise ValueError("Neat packing is incompatible with transformers>=4.53.0.")
_set_env_vars()
_verify_model_args(model_args, data_args, finetuning_args)
_check_extra_dependencies(model_args, finetuning_args, training_args)

3
src/llamafactory/model/adapter.py
View File

@ -16,11 +16,10 @@ import re
from typing import TYPE_CHECKING
import torch
from peft import LoraConfig, LoraModel, OFTConfig, OFTModel, PeftModel, TaskType, get_peft_model
from peft import LoraConfig, LoraModel, OFTConfig, PeftModel, TaskType, get_peft_model
from transformers.integrations import is_deepspeed_zero3_enabled
from ..extras import logging
from ..extras.misc import check_version
from .model_utils.misc import find_all_linear_modules, find_expanded_modules
from .model_utils.quantization import QuantizationMethod
from .model_utils.unsloth import get_unsloth_peft_model, load_unsloth_peft_model

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

@ -111,6 +111,7 @@ class CustomDPOTrainer(DPOTrainer):
if self.bco_gemma >= 1e-6:
from trl.trainer import RunningMoments
self.running = RunningMoments(self.accelerator)
@override
@ -161,7 +162,7 @@ class CustomDPOTrainer(DPOTrainer):
chosen_logps: "torch.Tensor",
rejected_logps: "torch.Tensor",
reference_chosen_logps: "torch.Tensor",
reference_rejected_logps: "torch.Tensor"
reference_rejected_logps: "torch.Tensor",
) -> "torch.Tensor":
chosen_logratios = chosen_logps - reference_chosen_logps
rejected_logratios = rejected_logps - reference_rejected_logps
@ -200,10 +201,7 @@ class CustomDPOTrainer(DPOTrainer):
if self.bco_gemma > 1e-6:
bco_losses = self.bco_loss(
policy_chosen_logps,
policy_rejected_logps,
reference_chosen_logps,
reference_rejected_logps
policy_chosen_logps, policy_rejected_logps, reference_chosen_logps, reference_rejected_logps
)
losses += bco_losses * self.bco_gemma
@ -288,7 +286,7 @@ class CustomDPOTrainer(DPOTrainer):
losses += self.ftx_gamma * sft_loss
if self.bco_gemma > 1e-6:
# re-weigthing for MPO
losses /= (self.ftx_gamma + self.bco_gemma + 1.0)
losses /= self.ftx_gamma + self.bco_gemma + 1.0
prefix = "eval_" if train_eval == "eval" else ""
metrics[f"{prefix}rewards/chosen"] = chosen_rewards.mean().item()