From 81b8a50aa55f8cbb7459dbdab1be5afcaeb04e6f Mon Sep 17 00:00:00 2001
From: jiaqiw09 <60021713+jiaqiw09@users.noreply.github.com>
Date: Sun, 4 Jan 2026 19:52:16 +0800
Subject: [PATCH] [deps] Update pyproject.toml and requirements (#9714)

Co-authored-by: Yaowei Zheng <hiyouga@buaa.edu.cn>
---
 .github/workflows/tests.yml                   |  3 +-
 .github/workflows/tests_cuda.yml              |  3 +-
 .github/workflows/tests_npu.yml               |  5 ++-
 README.md                                     | 40 ++++++-------------
 README_zh.md                                  | 39 ++++++------------
 docker/docker-cuda/Dockerfile                 |  3 +-
 docker/docker-cuda/Dockerfile.megatron        |  3 +-
 docker/docker-npu/Dockerfile                  |  3 +-
 docker/docker-rocm/Dockerfile                 |  3 +-
 pyproject.toml                                |  5 ---
 .../adam-mini.txt                             |  0
 .../requirements => requirements}/apollo.txt  |  0
 .../requirements => requirements}/aqlm.txt    |  0
 .../requirements => requirements}/badam.txt   |  0
 .../bitsandbytes.txt                          |  0
 requirements/deepspeed.txt                    |  1 +
 requirements/dev.txt                          |  4 ++
 .../requirements => requirements}/eetq.txt    |  0
 .../requirements => requirements}/fp8-te.txt  |  0
 .../requirements => requirements}/fp8.txt     |  0
 .../requirements => requirements}/galore.txt  |  0
 .../requirements => requirements}/gptq.txt    |  0
 .../requirements => requirements}/hqq.txt     |  0
 .../liger-kernel.txt                          |  0
 requirements/metrics.txt                      |  3 ++
 .../minicpm-v.txt                             |  0
 requirements/npu.txt                          |  4 ++
 .../openmind.txt                              |  0
 .../requirements => requirements}/sglang.txt  |  0
 .../requirements => requirements}/swanlab.txt |  0
 .../requirements => requirements}/vllm.txt    |  0
 scripts/eval_bleu_rouge.py                    |  2 +-
 32 files changed, 53 insertions(+), 68 deletions(-)
 rename {examples/requirements => requirements}/adam-mini.txt (100%)
 rename {examples/requirements => requirements}/apollo.txt (100%)
 rename {examples/requirements => requirements}/aqlm.txt (100%)
 rename {examples/requirements => requirements}/badam.txt (100%)
 rename {examples/requirements => requirements}/bitsandbytes.txt (100%)
 create mode 100644 requirements/deepspeed.txt
 create mode 100644 requirements/dev.txt
 rename {examples/requirements => requirements}/eetq.txt (100%)
 rename {examples/requirements => requirements}/fp8-te.txt (100%)
 rename {examples/requirements => requirements}/fp8.txt (100%)
 rename {examples/requirements => requirements}/galore.txt (100%)
 rename {examples/requirements => requirements}/gptq.txt (100%)
 rename {examples/requirements => requirements}/hqq.txt (100%)
 rename {examples/requirements => requirements}/liger-kernel.txt (100%)
 create mode 100644 requirements/metrics.txt
 rename {examples/requirements => requirements}/minicpm-v.txt (100%)
 create mode 100644 requirements/npu.txt
 rename {examples/requirements => requirements}/openmind.txt (100%)
 rename {examples/requirements => requirements}/sglang.txt (100%)
 rename {examples/requirements => requirements}/swanlab.txt (100%)
 rename {examples/requirements => requirements}/vllm.txt (100%)

diff --git a/.github/workflows/tests.yml b/.github/workflows/tests.yml
index f4166aab1..3d8f34dcd 100644
--- a/.github/workflows/tests.yml
+++ b/.github/workflows/tests.yml
@@ -70,7 +70,8 @@ jobs:
         run: |
           uv venv
           uv pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
-          uv pip install -e ".[dev]"
+          uv pip install -e .
+          uv pip install -r requirements/dev.txt
 
       - name: Install transformers
         if: ${{ matrix.transformers }}
diff --git a/.github/workflows/tests_cuda.yml b/.github/workflows/tests_cuda.yml
index 544936082..1c51e1c7a 100644
--- a/.github/workflows/tests_cuda.yml
+++ b/.github/workflows/tests_cuda.yml
@@ -52,7 +52,8 @@ jobs:
       - name: Install dependencies
         run: |
           uv venv
-          uv pip install -e ".[dev]"
+          uv pip install -e .
+          uv pip install -r requirements/dev.txt
 
       - name: Cache HuggingFace models
         id: hf-hub-cache
diff --git a/.github/workflows/tests_npu.yml b/.github/workflows/tests_npu.yml
index 8a4674f5c..da455babd 100644
--- a/.github/workflows/tests_npu.yml
+++ b/.github/workflows/tests_npu.yml
@@ -58,8 +58,9 @@ jobs:
       - name: Install dependencies
         run: |
           uv venv
-          uv pip install torch-npu==${{matrix.pytorch_npu}}
-          uv pip install -e ".[dev]"
+          uv pip install -r requirements/npu.txt
+          uv pip install -e .
+          uv pip install -r requirements/dev.txt
 
       - name: Install node
         run: |
diff --git a/README.md b/README.md
index ef9cc36fe..bf20e9090 100644
--- a/README.md
+++ b/README.md
@@ -517,10 +517,11 @@ huggingface-cli login
 ```bash
 git clone --depth 1 https://github.com/hiyouga/LLaMA-Factory.git
 cd LLaMA-Factory
-pip install -e ".[metrics]"
+pip install -e .
+pip install -r requirements/metrics.txt
 ```
 
-Optional dependencies available: `metrics`, `deepspeed`. Install with: `pip install -e ".[metrics,deepspeed]"`
+Optional dependencies available: `metrics`, `deepspeed`. Install with: `pip install -e . && pip install -r requirements/metrics.txt -r requirements/deepspeed.txt`
 
 Additional dependencies for specific features are available in `examples/requirements/`.
 
@@ -578,36 +579,21 @@ To enable FlashAttention-2 on the Windows platform, please use the script from [
 
 <details><summary>For Ascend NPU users</summary>
 
-To install LLaMA Factory on Ascend NPU devices, please upgrade Python to version 3.10 or higher: `pip install -e . torch-npu==2.7.1`. Additionally, you need to install the **[Ascend CANN Toolkit and Kernels](https://www.hiascend.com/developer/download/community/result?module=cann)**. Please follow the [installation tutorial](https://www.hiascend.com/document/detail/en/CANNCommunityEdition/600alphaX/softwareinstall/instg/atlasdeploy_03_0031.html) or use the following commands:
+To install LLaMA Factory on Ascend NPU devices, please upgrade Python to version 3.10 or higher: `pip install -r requirements/npu.txt`. Additionally, you need to install the **Ascend CANN Toolkit and Kernels**. Please follow the [installation tutorial](https://llamafactory.readthedocs.io/en/latest/advanced/npu_installation.html).
+
+
+You can also download the pre-built Docker images:
 
 ```bash
-# replace the url according to your CANN version and devices
-# install CANN Toolkit
-wget https://ascend-repo.obs.cn-east-2.myhuaweicloud.com/Milan-ASL/Milan-ASL%20V100R001C20SPC702/Ascend-cann-toolkit_8.0.0.alpha002_linux-"$(uname -i)".run
-bash Ascend-cann-toolkit_8.0.0.alpha002_linux-"$(uname -i)".run --install
+# Docker Hub
+docker pull hiyouga/llamafactory:latest-npu-a2
+docker pull hiyouga/llamafactory:latest-npu-a3
 
-# install CANN Kernels
-wget https://ascend-repo.obs.cn-east-2.myhuaweicloud.com/Milan-ASL/Milan-ASL%20V100R001C20SPC702/Ascend-cann-kernels-910b_8.0.0.alpha002_linux-"$(uname -i)".run
-bash Ascend-cann-kernels-910b_8.0.0.alpha002_linux-"$(uname -i)".run --install
-
-# set env variables
-source /usr/local/Ascend/ascend-toolkit/set_env.sh
+# quay.io
+docker pull quay.io/ascend/llamafactory:latest-npu-a2
+docker pull quay.io/ascend/llamafactory:latest-npu-a3
 ```
 
-| Requirement  | Minimum | Recommend      |
-| ------------ | ------- | -------------- |
-| CANN         | 8.0.RC1 | 8.0.0.alpha002 |
-| torch        | 2.1.0   | 2.7.1          |
-| torch-npu    | 2.1.0   | 2.7.1          |
-| deepspeed    | 0.13.2  | 0.13.2         |
-| vllm-ascend  | -       | 0.7.3          |
-
-Remember to use `ASCEND_RT_VISIBLE_DEVICES` instead of `CUDA_VISIBLE_DEVICES` to specify the device to use.
-
-If you cannot infer model on NPU devices, try setting `do_sample: false` in the configurations.
-
-Download the pre-built Docker images: [32GB](http://mirrors.cn-central-221.ovaijisuan.com/detail/130.html) | [64GB](http://mirrors.cn-central-221.ovaijisuan.com/detail/131.html)
-
 #### Install BitsAndBytes
 
 To use QLoRA based on bitsandbytes on Ascend NPU, please follow these 3 steps:
diff --git a/README_zh.md b/README_zh.md
index 2fad20d3c..d2f543f13 100644
--- a/README_zh.md
+++ b/README_zh.md
@@ -519,10 +519,11 @@ huggingface-cli login
 ```bash
 git clone --depth 1 https://github.com/hiyouga/LLaMA-Factory.git
 cd LLaMA-Factory
-pip install -e ".[metrics]"
+pip install -e .
+pip install -r requirements/metrics.txt
 ```
 
-可选的额外依赖项：`metrics`、`deepspeed`。使用 `pip install -e ".[metrics,deepspeed]"` 安装。
+可选的额外依赖项：`metrics`、`deepspeed`。使用 `pip install -e . && pip install -r requirements/metrics.txt -r requirements/deepspeed.txt` 安装。
 
 其他可选依赖项请参考 `examples/requirements/` 目录下的文件。
 
@@ -580,36 +581,20 @@ pip install https://github.com/jllllll/bitsandbytes-windows-webui/releases/downl
 
 <details><summary>昇腾 NPU 用户指南</summary>
 
-在昇腾 NPU 设备上安装 LLaMA Factory 时，请升级 Python 到 3.10 及以上，并需要指定额外依赖项，使用 `pip install -e . torch-npu==2.7.1` 命令安装。此外，还需要安装 **[Ascend CANN Toolkit 与 Kernels](https://www.hiascend.com/developer/download/community/result?module=cann)**，安装方法请参考[安装教程](https://www.hiascend.com/document/detail/zh/CANNCommunityEdition/80RC2alpha002/quickstart/quickstart/quickstart_18_0004.html)或使用以下命令：
+在昇腾 NPU 设备上安装 LLaMA Factory 时，请升级 Python 到 3.10 及以上，并需要指定额外依赖项，使用 `pip install -r requirements/npu.txt` 命令安装。此外，还需要安装 **Ascend CANN Toolkit 与 Kernels**，安装方法请参考[安装教程](https://llamafactory.readthedocs.io/zh-cn/latest/advanced/npu_installation.html)。
+
+您可以直接下载预安装的最新docker镜像：
 
 ```bash
-# 请替换 URL 为 CANN 版本和设备型号对应的 URL
-# 安装 CANN Toolkit
-wget https://ascend-repo.obs.cn-east-2.myhuaweicloud.com/Milan-ASL/Milan-ASL%20V100R001C17SPC701/Ascend-cann-toolkit_8.0.RC1.alpha001_linux-"$(uname -i)".run
-bash Ascend-cann-toolkit_8.0.RC1.alpha001_linux-"$(uname -i)".run --install
+# Docker Hub
+docker pull hiyouga/llamafactory:latest-npu-a2
+docker pull hiyouga/llamafactory:latest-npu-a3
 
-# 安装 CANN Kernels
-wget https://ascend-repo.obs.cn-east-2.myhuaweicloud.com/Milan-ASL/Milan-ASL%20V100R001C17SPC701/Ascend-cann-kernels-910b_8.0.RC1.alpha001_linux.run
-bash Ascend-cann-kernels-910b_8.0.RC1.alpha001_linux.run --install
-
-# 设置环境变量
-source /usr/local/Ascend/ascend-toolkit/set_env.sh
+# quay.io
+docker pull quay.io/ascend/llamafactory:latest-npu-a2
+docker pull quay.io/ascend/llamafactory:latest-npu-a3
 ```
 
-| 依赖项        | 至少     | 推荐           |
-| ------------ | ------- | -------------- |
-| CANN         | 8.0.RC1 | 8.0.0.alpha002 |
-| torch        | 2.1.0   | 2.7.1          |
-| torch-npu    | 2.1.0   | 2.7.1          |
-| deepspeed    | 0.13.2  | 0.13.2         |
-| vllm-ascend  | -       | 0.7.3          |
-
-请使用 `ASCEND_RT_VISIBLE_DEVICES` 而非 `CUDA_VISIBLE_DEVICES` 来指定运算设备。
-
-如果遇到无法正常推理的情况，请尝试设置 `do_sample: false`。
-
-下载预构建 Docker 镜像：[32GB](http://mirrors.cn-central-221.ovaijisuan.com/detail/130.html) | [64GB](http://mirrors.cn-central-221.ovaijisuan.com/detail/131.html)
-
 #### 安装 BitsAndBytes
 
 如果要在 Ascend NPU 上进行基于 bitsandbytes 的 QLoRA 量化微调，请执行如下步骤：
diff --git a/docker/docker-cuda/Dockerfile b/docker/docker-cuda/Dockerfile
index b70ad5b50..f33d14e33 100644
--- a/docker/docker-cuda/Dockerfile
+++ b/docker/docker-cuda/Dockerfile
@@ -32,7 +32,8 @@ RUN pip config set global.index-url "${PIP_INDEX}" && \
 COPY . /app
 
 # Install LLaMA Factory
-RUN pip install --no-cache-dir --no-build-isolation -e ".[metrics,deepspeed]"
+RUN pip install --no-cache-dir --no-build-isolation -e . && \
+    pip install --no-cache-dir --no-build-isolation -r requirements/metrics.txt -r requirements/deepspeed.txt
 
 # Rebuild flash attention
 RUN if [ "${INSTALL_FLASHATTN}" == "true" ]; then \
diff --git a/docker/docker-cuda/Dockerfile.megatron b/docker/docker-cuda/Dockerfile.megatron
index 7f457fefd..8b44434d4 100644
--- a/docker/docker-cuda/Dockerfile.megatron
+++ b/docker/docker-cuda/Dockerfile.megatron
@@ -60,7 +60,8 @@ WORKDIR /app
 COPY . /app
 
 # Install LLaMA Factory
-RUN pip install --no-cache-dir -e ".[metrics]" --no-build-isolation
+RUN pip install --no-cache-dir -e . --no-build-isolation && \
+    pip install --no-cache-dir -r requirements/metrics.txt --no-build-isolation
 
 RUN pip install "git+https://github.com/alibaba/roll.git#subdirectory=mcore_adapter"
 
diff --git a/docker/docker-npu/Dockerfile b/docker/docker-npu/Dockerfile
index d3c7acbaa..364774f85 100644
--- a/docker/docker-npu/Dockerfile
+++ b/docker/docker-npu/Dockerfile
@@ -35,7 +35,8 @@ COPY . /app
 # Install torch-npu
 RUN pip uninstall -y torch torchvision torchaudio && \
     pip install --no-cache-dir "torch==2.7.1" "torch-npu==2.7.1" "torchvision==0.22.1" "torchaudio==2.7.1" --index-url "${PYTORCH_INDEX}" && \
-    pip install --no-cache-dir -e ".[metrics]" --no-build-isolation
+    pip install --no-cache-dir -e . --no-build-isolation && \
+    pip install --no-cache-dir -r requirements/metrics.txt --no-build-isolation
 
 # Set up volumes
 # VOLUME [ "/root/.cache/huggingface", "/app/shared_data", "/app/output" ]
diff --git a/docker/docker-rocm/Dockerfile b/docker/docker-rocm/Dockerfile
index d0a981767..4e82be572 100644
--- a/docker/docker-rocm/Dockerfile
+++ b/docker/docker-rocm/Dockerfile
@@ -34,7 +34,8 @@ COPY . /app
 
 # Reinstall pytorch rocm and install LLaMA Factory
 RUN pip uninstall -y torch torchvision torchaudio && \
-    pip install --no-cache-dir --no-build-isolation -e --pre ".[metrics,deepspeed]" --index-url "${PYTORCH_INDEX}"
+    pip install --no-cache-dir --no-build-isolation -e --pre . --index-url "${PYTORCH_INDEX}" && \
+    pip install --no-cache-dir --no-build-isolation -r requirements/metrics.txt -r requirements/deepspeed.txt --index-url "${PYTORCH_INDEX}"
 
 # Rebuild flash attention
 RUN if [ "${INSTALL_FLASHATTN}" == "true" ]; then \
diff --git a/pyproject.toml b/pyproject.toml
index ef03b82f8..357df9314 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -76,11 +76,6 @@ dependencies = [
     "sse-starlette"
 ]
 
-[project.optional-dependencies]
-dev = ["pre-commit", "ruff", "pytest", "build"]
-metrics = ["nltk", "jieba", "rouge-chinese"]
-deepspeed = ["deepspeed>=0.10.0,<=0.16.9"]
-
 [project.scripts]
 llamafactory-cli = "llamafactory.cli:main"
 lmf = "llamafactory.cli:main"
diff --git a/examples/requirements/adam-mini.txt b/requirements/adam-mini.txt
similarity index 100%
rename from examples/requirements/adam-mini.txt
rename to requirements/adam-mini.txt
diff --git a/examples/requirements/apollo.txt b/requirements/apollo.txt
similarity index 100%
rename from examples/requirements/apollo.txt
rename to requirements/apollo.txt
diff --git a/examples/requirements/aqlm.txt b/requirements/aqlm.txt
similarity index 100%
rename from examples/requirements/aqlm.txt
rename to requirements/aqlm.txt
diff --git a/examples/requirements/badam.txt b/requirements/badam.txt
similarity index 100%
rename from examples/requirements/badam.txt
rename to requirements/badam.txt
diff --git a/examples/requirements/bitsandbytes.txt b/requirements/bitsandbytes.txt
similarity index 100%
rename from examples/requirements/bitsandbytes.txt
rename to requirements/bitsandbytes.txt
diff --git a/requirements/deepspeed.txt b/requirements/deepspeed.txt
new file mode 100644
index 000000000..fe9116909
--- /dev/null
+++ b/requirements/deepspeed.txt
@@ -0,0 +1 @@
+deepspeed>=0.10.0,<=0.16.9
diff --git a/requirements/dev.txt b/requirements/dev.txt
new file mode 100644
index 000000000..f0fc17fcc
--- /dev/null
+++ b/requirements/dev.txt
@@ -0,0 +1,4 @@
+pre-commit
+ruff
+pytest
+build
diff --git a/examples/requirements/eetq.txt b/requirements/eetq.txt
similarity index 100%
rename from examples/requirements/eetq.txt
rename to requirements/eetq.txt
diff --git a/examples/requirements/fp8-te.txt b/requirements/fp8-te.txt
similarity index 100%
rename from examples/requirements/fp8-te.txt
rename to requirements/fp8-te.txt
diff --git a/examples/requirements/fp8.txt b/requirements/fp8.txt
similarity index 100%
rename from examples/requirements/fp8.txt
rename to requirements/fp8.txt
diff --git a/examples/requirements/galore.txt b/requirements/galore.txt
similarity index 100%
rename from examples/requirements/galore.txt
rename to requirements/galore.txt
diff --git a/examples/requirements/gptq.txt b/requirements/gptq.txt
similarity index 100%
rename from examples/requirements/gptq.txt
rename to requirements/gptq.txt
diff --git a/examples/requirements/hqq.txt b/requirements/hqq.txt
similarity index 100%
rename from examples/requirements/hqq.txt
rename to requirements/hqq.txt
diff --git a/examples/requirements/liger-kernel.txt b/requirements/liger-kernel.txt
similarity index 100%
rename from examples/requirements/liger-kernel.txt
rename to requirements/liger-kernel.txt
diff --git a/requirements/metrics.txt b/requirements/metrics.txt
new file mode 100644
index 000000000..533493d82
--- /dev/null
+++ b/requirements/metrics.txt
@@ -0,0 +1,3 @@
+nltk
+jieba
+rouge-chinese
diff --git a/examples/requirements/minicpm-v.txt b/requirements/minicpm-v.txt
similarity index 100%
rename from examples/requirements/minicpm-v.txt
rename to requirements/minicpm-v.txt
diff --git a/requirements/npu.txt b/requirements/npu.txt
new file mode 100644
index 000000000..054bd93ea
--- /dev/null
+++ b/requirements/npu.txt
@@ -0,0 +1,4 @@
+torch==2.7.1
+torch-npu==2.7.1
+torchvision==0.22.1
+torchaudio==2.7.1
diff --git a/examples/requirements/openmind.txt b/requirements/openmind.txt
similarity index 100%
rename from examples/requirements/openmind.txt
rename to requirements/openmind.txt
diff --git a/examples/requirements/sglang.txt b/requirements/sglang.txt
similarity index 100%
rename from examples/requirements/sglang.txt
rename to requirements/sglang.txt
diff --git a/examples/requirements/swanlab.txt b/requirements/swanlab.txt
similarity index 100%
rename from examples/requirements/swanlab.txt
rename to requirements/swanlab.txt
diff --git a/examples/requirements/vllm.txt b/requirements/vllm.txt
similarity index 100%
rename from examples/requirements/vllm.txt
rename to requirements/vllm.txt
diff --git a/scripts/eval_bleu_rouge.py b/scripts/eval_bleu_rouge.py
index 22e370bc7..4ff96dd8e 100644
--- a/scripts/eval_bleu_rouge.py
+++ b/scripts/eval_bleu_rouge.py
@@ -28,7 +28,7 @@ try:
     jieba.setLogLevel(logging.CRITICAL)
     jieba.initialize()
 except ImportError:
-    print("Please install llamafactory with `pip install -e .[metrics]`.")
+    print("Please install llamafactory with `pip install -r requirements/metrics.txt`.")
     raise