Skip to content

Update to newer pytorch, trtllm, and vllm #548

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
chtruong814 wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

Update to newer pytorch, trtllm, and vllm #548

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
82165c9
Update to newer pytorch, trtllm, and vllm
chtruong814 Jan 9, 2026
0d2c536
Add nemo lightning extra
chtruong814 Jan 11, 2026
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
2 changes: 1 addition & 1 deletion docker/Dockerfile.pytorch
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.

FROM nvcr.io/nvidia/pytorch:25.06-py3
FROM nvcr.io/nvidia/pytorch:25.10-py3
WORKDIR /workspace
COPY . .
ARG INFERENCE_FRAMEWORK
Expand Down
30 changes: 15 additions & 15 deletions docker/common/install.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,12 +82,12 @@ main() {
dpkg -i cuda-keyring_1.1-1_all.deb
rm cuda-keyring_1.1-1_all.deb
apt-get update
apt-get install -y cuda-toolkit-12-8 cudnn-cuda-12 libcudnn9-cuda-12 libcutlass-dev
apt-get install -y cuda-toolkit-12-8 cudnn-cuda-12 libcudnn9-cuda-12 libcutlass-dev

# Install OpenMPI and nvinfer
apt-get install -y libopenmpi-dev libnvinfer-lean-dev
fi

# Clean up
apt-get clean

Expand All @@ -109,18 +109,18 @@ main() {
"--no-install-package" "torch"
"--no-install-package" "torchvision"
"--no-install-package" "triton"
"--no-install-package" "nvidia-cublas-cu12"
"--no-install-package" "nvidia-cuda-cupti-cu12"
"--no-install-package" "nvidia-cuda-nvrtc-cu12"
"--no-install-package" "nvidia-cuda-runtime-cu12"
"--no-install-package" "nvidia-cudnn-cu12"
"--no-install-package" "nvidia-cufft-cu12"
"--no-install-package" "nvidia-cufile-cu12"
"--no-install-package" "nvidia-curand-cu12"
"--no-install-package" "nvidia-cusolver-cu12"
"--no-install-package" "nvidia-cusparse-cu12"
"--no-install-package" "nvidia-cusparselt-cu12"
"--no-install-package" "nvidia-nccl-cu12"
"--no-install-package" "nvidia-cublas"
"--no-install-package" "nvidia-cuda-cupti"
"--no-install-package" "nvidia-cuda-nvrtc"
"--no-install-package" "nvidia-cuda-runtime"
"--no-install-package" "nvidia-cudnn-cu13"
"--no-install-package" "nvidia-cufft"
"--no-install-package" "nvidia-cufile"
"--no-install-package" "nvidia-curand"
"--no-install-package" "nvidia-cusolver"
"--no-install-package" "nvidia-cusparse"
"--no-install-package" "nvidia-cusparselt-cu13"
"--no-install-package" "nvidia-nccl-cu13"
)
else
UV_ARGS=()
Expand Down Expand Up @@ -152,7 +152,7 @@ main() {
pip install --pre --no-cache-dir --upgrade pip
pip install --pre --no-cache-dir 'torch>=2.7.0,<2.8.0' pybind11 wheel_stub ninja wheel packaging


pip install --pre --no-cache-dir --no-build-isolation .$EXTRA
fi

Expand Down
26 changes: 14 additions & 12 deletions pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,7 +43,7 @@ license = { text = "Apache 2.0" }
dependencies = [
"megatron-core>=0.14.0a0,<0.15.0",
"megatron-bridge>=0.1.0a0,<0.2.0",
"nvidia-modelopt[torch]>=0.33.0a0,<0.34.0; sys_platform != 'darwin'",
"nvidia-modelopt[torch]>=0.37.0a0,<0.41.0; sys_platform != 'darwin'",
"nvidia-resiliency-ext>=0.4.0a0,<0.5.0; sys_platform != 'darwin'",
"transformer-engine[pytorch]>=2.6.0a0,<2.7.0; sys_platform != 'darwin'",
"accelerate",
Expand All @@ -61,7 +61,7 @@ dependencies = [
"lightning",
"omegaconf>=2.3.0",
"peft<0.14.0",
"torch==2.7.1",
"torch==2.9.0",
"torchvision",
"torchmetrics>=0.11.0",
"wandb",
Expand All @@ -80,13 +80,13 @@ dependencies = [

[project.optional-dependencies]
inframework = []
vllm = ["vllm~=0.10.0", "pandas", "timm"]
trtllm = ["tensorrt-llm>=1.0.0a0,<1.1.0,>=1.0.0rc6", "cuda-python~=12.8.0"]
trt-onnx = ["tensorrt==10.11.0.33", "onnx==1.18.0", "transformers==4.51.3"]
vllm = ["vllm~=0.13.0", "pandas", "timm"]
trtllm = ["tensorrt-llm>=1.1.0a0,<1.2.0", "cuda-python~=13.0.0"]
trt-onnx = ["tensorrt==10.14.1.48", "onnx==1.18.0", "transformers==4.51.3"]

[dependency-groups]
# This is a default group so that we install these even with bare `uv sync`
build = ["setuptools", "torch==2.7.1", "pybind11", "Cython>=3.0.0", "ninja"]
build = ["setuptools", "torch==2.9.0", "pybind11", "Cython>=3.0.0", "ninja"]
docs = [
"sphinx",
"sphinx-autobuild", # For live doc serving while editing docs
Expand All @@ -98,15 +98,15 @@ docs = [
linting = ["pre-commit>=3.6.0", "ruff~=0.9.0"]
test = ["pytest", "pytest-mock", "coverage", "click"]
nemo-toolkit = [
"nemo-toolkit[automodel,common-only,nlp-only,eval,multimodal-only]>=2.5.0a0,<2.6.0",
"nemo-toolkit[automodel,common-only,nlp-only,eval,multimodal-only,lightning]>=2.6.0a0,<2.8.0",
]
nemo-run = ["nemo-run"]

[tool.uv.sources]
xformers = [{ index = "pytorch-cu128" }]
torch = [{ index = "pytorch-cu128" }]
xformers = [{ index = "pytorch-cu130" }]
torch = [{ index = "pytorch-cu130" }]
vllm = [
{ index = "pytorch-cu128", marker = "python_version < '3.9' and platform_machine == 'x86_64'" },
{ index = "pytorch-cu130", marker = "python_version < '3.9' and platform_machine == 'x86_64'" },
{ index = "pypi", marker = "platform_machine == 'aarch64'" },
]
# megatron-bridge = { git = "https://github.com/NVIDIA-NeMo/Megatron-Bridge.git", rev = "ecf05926b4765aada82c8eabab4a374e8e83a9c5" }
Expand All @@ -130,6 +130,8 @@ override-dependencies = [
"urllib3>1.27.0",
"tiktoken>=0.9.0", # because nemo-toolkit and megatron-bridge disagree on tiktoken, we need to pin it here,
"fsspec[http]>=2023.1.0,<=2024.9.0",
"patchelf; sys_platform == 'never'",
"transformers==4.51.3"
]
prerelease = "allow"

Expand All @@ -139,8 +141,8 @@ url = "https://pypi.org/simple"
explicit = true

[[tool.uv.index]]
name = "pytorch-cu128"
url = "https://download.pytorch.org/whl/cu128"
name = "pytorch-cu130"
url = "https://download.pytorch.org/whl/cu130"
explicit = true

[tool.pytest.ini_options]
Expand Down
Loading
Loading