Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
121 changes: 0 additions & 121 deletions .github/workflows/cuda12.8_whl_release.yml

This file was deleted.

8 changes: 4 additions & 4 deletions .github/workflows/pypi.yml
Original file line number Diff line number Diff line change
Expand Up @@ -18,8 +18,8 @@ jobs:
env:
PYTHON_VERSION: ${{ matrix.pyver }}
PLAT_NAME: manylinux2014_x86_64
DOCKER_TAG: cuda12.4
OUTPUT_FOLDER: cuda12_dist
DOCKER_TAG: cuda12.8
OUTPUT_FOLDER: cuda12.8_dist
steps:
- name: Free disk space
uses: jlumbroso/free-disk-space@main
Expand Down Expand Up @@ -75,11 +75,11 @@ jobs:
shell: pwsh
run: ./builder/windows/setup_cuda.ps1
env:
INPUT_CUDA_VERSION: '12.6.2'
INPUT_CUDA_VERSION: '12.8.1'
- name: Build wheel
run: |
python -m build --wheel -o build/wheel
Get-ChildItem -Path "build" -Filter "*.whl" | ForEach-Object { change_wheel_version $_.FullName --local-version cu121 --delete-old-wheel }
Get-ChildItem -Path "build" -Filter "*.whl" | ForEach-Object { change_wheel_version $_.FullName --local-version cu128 --delete-old-wheel }
Copy link

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In the Windows build step, python -m build --wheel -o build/wheel writes wheels under build/wheel, but the subsequent Get-ChildItem -Path "build" -Filter "*.whl" won’t find them (non-recursive), so change_wheel_version likely never runs. Point Get-ChildItem at build/wheel (or use -Recurse) so the local version update is actually applied (or remove the step if it’s not needed).

Suggested change
Get-ChildItem -Path "build" -Filter "*.whl" | ForEach-Object { change_wheel_version $_.FullName --local-version cu128 --delete-old-wheel }
Get-ChildItem -Path "build/wheel" -Filter "*.whl" | ForEach-Object { change_wheel_version $_.FullName --local-version cu128 --delete-old-wheel }

Copilot uses AI. Check for mistakes.
- name: Upload Artifacts
uses: actions/upload-artifact@v4
with:
Expand Down
10 changes: 1 addition & 9 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -224,15 +224,7 @@ conda activate lmdeploy
pip install lmdeploy
```

Since v0.3.0, the default prebuilt package is compiled on **CUDA 12**. Starting from v0.10.2, LMDeploy no longer supports CUDA 11 series.

If you are using a GeForce RTX 50 series graphics card, please install the LMDeploy prebuilt package compiled with **CUDA 12.8** as follows:

```shell
export LMDEPLOY_VERSION=0.12.3
export PYTHON_VERSION=312
pip install https://github.com/InternLM/lmdeploy/releases/download/v${LMDEPLOY_VERSION}/lmdeploy-${LMDEPLOY_VERSION}+cu128-cp${PYTHON_VERSION}-cp${PYTHON_VERSION}-manylinux2014_x86_64.whl --extra-index-url https://download.pytorch.org/whl/cu128
```
Starting from **v0.13.0**, the default prebuilt wheels published on **PyPI** are built against **CUDA 12.8**, so `pip install lmdeploy` is sufficient for typical setups including GeForce RTX 50 series.

## Offline Batch Inference

Expand Down
5 changes: 4 additions & 1 deletion README_ja.md
Original file line number Diff line number Diff line change
Expand Up @@ -201,7 +201,10 @@ conda activate lmdeploy
pip install lmdeploy
```

v0.3.0から、デフォルトの事前構築済みパッケージはCUDA 12でコンパイルされています。
**v0.13.0** 以降、**PyPI** に公開される既定の事前構築wheelは **CUDA 12.8** 向けにビルドされています。v0.10.2以降、LMDeployはCUDA 11系をサポートしません。

GeForce RTX 50シリーズを含む一般的な用途でも、上記の `pip install lmdeploy` で問題ありません。

CUDA 11+プラットフォームでのインストールに関する情報、またはソースからのビルド手順については、[インストールガイドを](docs/en/get_started/installation.md)参照してください。

## オフラインバッチ推論
Expand Down
10 changes: 1 addition & 9 deletions README_zh-CN.md
Original file line number Diff line number Diff line change
Expand Up @@ -226,15 +226,7 @@ conda activate lmdeploy
pip install lmdeploy
```

自 v0.3.0 版本起,默认预编译包基于 **CUDA 12** 编译。v0.10.2 及更高版本中,LMDeploy 不再支持 CUDA 11+。

若使用 GeForce RTX 50 系列显卡,请按照如下方式安装基于 **CUDA 12.8** 编译的 LMDeploy 预编译包。

```shell
export LMDEPLOY_VERSION=0.12.3
export PYTHON_VERSION=312
pip install https://github.com/InternLM/lmdeploy/releases/download/v${LMDEPLOY_VERSION}/lmdeploy-${LMDEPLOY_VERSION}+cu128-cp${PYTHON_VERSION}-cp${PYTHON_VERSION}-manylinux2014_x86_64.whl --extra-index-url https://download.pytorch.org/whl/cu128
```
自 **v0.13.0** 起,**PyPI** 上默认预编译 wheel 基于 **CUDA 12.8** 构建,一般用户(含 GeForce RTX 50 系列)使用上方的 `pip install lmdeploy` 即可。

## 离线批处理

Expand Down
2 changes: 1 addition & 1 deletion lmdeploy/version.py
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
# Copyright (c) OpenMMLab. All rights reserved.

__version__ = '0.12.3'
__version__ = '0.13.0'
short_version = __version__


Expand Down
Loading