Skip to content

Revert "[None][infra] Bump black and tornado (#12876)" #12894

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

Merged
yuanjingx87 merged 2 commits into from
Apr 9, 2026
+8 −11
Merged

Revert "[None][infra] Bump black and tornado (#12876)" #12894

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
32c65ac
Revert "[None][infra] Bump black and tornado (#12876)"
niukuo Apr 9, 2026
44a6e22
[None][infra] update docker images
niukuo Apr 9, 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
4 changes: 0 additions & 4 deletions constraints.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,3 @@ wheel>=0.46.2
protobuf>=6.33.5
# WAR against https://github.com/advisories/GHSA-6mq8-rvhq-8wgg
aiohttp>=3.13.3
# WAR against https://github.com/advisories/GHSA-qjxf-f2mg-c6mc
tornado>=6.5.5
# WAR against https://github.com/advisories/GHSA-3936-cmfr-pm3m
black>=26.3.1
1 change: 1 addition & 0 deletions examples/models/core/gemma/requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
nvidia-cudnn-cu12~=8.9; platform_machine == "x86_64"
tensorrt_llm>=0.0.0.dev0
flax~=0.8.0
numpy<2
# jax[cuda12_pip]~=0.4.19
safetensors~=0.4.1
sentencepiece>=0.1.99
Expand Down
8 changes: 4 additions & 4 deletions jenkins/current_image_tags.properties
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@
# images are adopted from PostMerge pipelines, the abbreviated commit hash is used instead.
IMAGE_NAME=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm

LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.12-py3-x86_64-ubuntu24.04-trt10.14.1.48-skip-tritondevel-202603011156-11778
LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.12-py3-aarch64-ubuntu24.04-trt10.14.1.48-skip-tritondevel-202603011156-11778
LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-13.1.0-devel-rocky8-x86_64-rocky8-py310-trt10.14.1.48-skip-tritondevel-202603011156-11778
LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-13.1.0-devel-rocky8-x86_64-rocky8-py312-trt10.14.1.48-skip-tritondevel-202603011156-11778
LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:x86_64-tritondevel-torch_skip-cc469c6-github-pr-12894-1854
LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:sbsa-tritondevel-torch_skip-cc469c6-github-pr-12894-1854
LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:x86_64-rockylinux8-torch_skip-py310-cc469c6-github-pr-12894-1854
LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm-staging/tritondevel:x86_64-rockylinux8-torch_skip-py312-cc469c6-github-pr-12894-1854
2 changes: 1 addition & 1 deletion requirements-dev.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,5 +36,5 @@ opentelemetry-api>=1.26.0
opentelemetry-exporter-otlp>=1.26.0
opentelemetry-semantic-conventions-ai>=0.4.1
fuzzywuzzy==0.18.0
aiperf==0.6.0
aiperf==0.4.0
nanobind>=2.9.0
4 changes: 2 additions & 2 deletions requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ cuda-python>=13
diffusers>=0.27.0
lark
mpi4py
numpy>=2.0.0,<2.4 # numba 0.63.1 requires numpy<2.4
numpy<2
onnx>=1.18.0,<1.20.0
onnx_graphsurgeon>=0.5.2
openai
Expand Down Expand Up @@ -70,7 +70,7 @@ blobfile
openai-harmony==0.0.4
nvidia-cutlass-dsl==4.3.4; python_version >= "3.10"
plotly
numexpr
numexpr<2.14.0 # WAR for attempted use of nonexistent numpy.typing
Copy link
Copy Markdown
Contributor

@coderabbitai coderabbitai bot Apr 9, 2026

Choose a reason for hiding this comment

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

⚠️ Potential issue | 🟠 Major

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Read-only verification: inspect current numexpr usage and related comments.
set -euo pipefail

echo "== numexpr references in requirements/constraints =="
rg -n -C2 'numexpr|numpy\.typing' --glob '**/*requirements*.txt' --glob '**/*constraints*.txt'

echo
echo "== Python imports/usages =="
rg -n -C3 '\bimport numexpr\b|\bnumexpr\.' --type=py

Repository: NVIDIA/TensorRT-LLM

Length of output: 463

🌐 Web query:

Is there a documented numexpr issue or release note in/after 2.14.0 related to numpy.typing compatibility, and what is the recommended compatible version range with NumPy 1.26.x?

💡 Result:

Yes.

  • Documented issue / release note: NumExpr 2.14.0 introduced “static typing support” (it imports/uses numpy.typing).[1] In 2.14.1, the release notes explicitly say static typing support was rolled back “to ensure compatibility with NumPy 1.26” and that CI tests for NumPy 1.26 were added.[2]
  • Recommended NumExpr version range for NumPy 1.26.x: use numexpr ≥ 2.14.1 (and avoid 2.14.0). NumExpr 2.14.1 also sets its NumPy requirement to numpy >= 1.26.0.[3]

Sources: [1] NumExpr 2.14.0 release notes, [2] NumExpr 2.14.1 release notes, [3] NumExpr 2.14.1 requirements.txt.

Reconsider numexpr constraint; 2.14.1+ resolves the documented issue.

The constraint numexpr<2.14.0 correctly avoids the broken 2.14.0 release but blocks fixed versions. NumExpr 2.14.0 introduced numpy.typing support (causing compatibility failures with NumPy 1.26), and 2.14.1 explicitly rolled back this feature per its release notes.

Either:

  • Update to allow the fixed version: numexpr>=2.14.1 (preferred if compatible), or
  • Explicitly exclude only the broken version: numexpr!=2.14.0 plus a reasonable lower bound.

Document the constraint with a reference: # WAR: numexpr 2.14.0 incompatible with numpy.typing; fixed in 2.14.1 (https://github.com/pydata/numexpr/releases/tag/v2.14.1)

🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In `@requirements.txt` at line 73, The dependency line restricting numexpr to
"<2.14.0" is too broad and blocks the patched release; update the requirement
for the package symbol "numexpr" to allow the fixed version by replacing the
constraint with either "numexpr>=2.14.1" (preferred) or "numexpr!=2.14.0,<X.Y.Z"
if you want an upper bound, and add an inline comment referencing the issue/fix
(e.g., "# WAR: numexpr 2.14.0 incompatible with numpy.typing; fixed in 2.14.1
(https://github.com/pydata/numexpr/releases/tag/v2.14.1)") so the rationale is
preserved.

partial_json_parser
apache-tvm-ffi==0.1.6 # used for reduce nvidia-cutlass-dsl host overhead
torch-c-dlpack-ext==0.1.3 # used for reduce nvidia-cutlass-dsl host overhead, optional package for improved torch tensor calling perf
Expand Down
Loading