Skip to content

[deps][LLM] Upgrade vLLM to 0.15.0 #60253

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

Closed
nrghosh wants to merge 20 commits into from
Closed

[deps][LLM] Upgrade vLLM to 0.15.0 #60253

Show file tree
Hide file tree
Changes from 9 commits
Commits
Show all changes
20 commits
Select commit Hold shift + click to select a range
cd095ee
[LLM] Point vLLM dependency to v0.14.0rc1 for upgrade testing
nrghosh Jan 17, 2026
758b2c9
overriding vllm torch requirement
elliot-barn Jan 18, 2026
923dabc
including torch and vllm
elliot-barn Jan 19, 2026
fc6b087
[LLM] Update vLLM from rc1 to 0.14.0 release
nrghosh Jan 21, 2026
c1c11e7
wip - compile deps
nrghosh Jan 21, 2026
d18c71c
wip - fix test - PoolingParams.normalize deprecated in favor of use_a…
nrghosh Jan 21, 2026
52c45be
Fix multi-gpu llm tests
nrghosh Jan 22, 2026
3190ad1
use smaller MoE model to avoid OOms in CI
nrghosh Jan 22, 2026
eca9898
Upgrade to vLLM 0.15.0 take 2
jeffreywang88 Jan 30, 2026
24e1d99
Take 3
jeffreywang88 Jan 30, 2026
231de88
Remove extra-index-url
jeffreywang88 Jan 30, 2026
04eb5d2
Attempt to fix CUDA version incompatibility
jeffreywang88 Jan 30, 2026
4cd9ecc
OpenAI protocol adjustment
jeffreywang88 Jan 30, 2026
0d9957c
linter
jeffreywang88 Jan 30, 2026
b749a9f
Trim
jeffreywang88 Jan 30, 2026
9f9958c
Run bazel run //ci/raydepsets:raydepsets -- build --all-configs --che…
jeffreywang88 Jan 30, 2026
8248ad7
Use parse_chat_messages_async
jeffreywang88 Jan 30, 2026
6711055
Clean up
jeffreywang88 Jan 31, 2026
cda414a
Recompile locally
jeffreywang88 Jan 31, 2026
3ae3f3e
Clean up
jeffreywang88 Feb 1, 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: 2 additions & 2 deletions ci/compile_llm_requirements.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,9 +6,9 @@ CONFIG_PATH="${1:-ci/raydepsets/configs/rayllm.depsets.yaml}"

mkdir -p /tmp/ray-deps

# Remove the GPU constraints
# Remove the GPU constraints, numpy pin, and scipy pin (LLM requires numpy>=2 and compatible scipy)
cp python/requirements_compiled.txt /tmp/ray-deps/requirements_compiled.txt
sed -e '/^--extra-index-url /d' -e '/^--find-links /d' /tmp/ray-deps/requirements_compiled.txt > /tmp/ray-deps/requirements_compiled.txt.tmp
sed -e '/^--extra-index-url /d' -e '/^--find-links /d' -e '/^numpy==/d' -e '/^scipy==/d' /tmp/ray-deps/requirements_compiled.txt > /tmp/ray-deps/requirements_compiled.txt.tmp
mv /tmp/ray-deps/requirements_compiled.txt.tmp /tmp/ray-deps/requirements_compiled.txt

bazel run //ci/raydepsets:raydepsets -- build "${CONFIG_PATH}"
Expand Down
1 change: 1 addition & 0 deletions ci/raydepsets/configs/llm_release_tests.depsets.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@ build_arg_sets:
- --unsafe-package ray
- --python-platform=linux
- --extra-index-url https://download.pytorch.org/whl/${CUDA_CODE}
- --override python/requirements/llm/torch-override.txt
build_arg_sets:
- cu128

Expand Down
5 changes: 3 additions & 2 deletions ci/raydepsets/configs/rayimg.depsets.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,8 +39,8 @@ depsets:
- release/ray_release/byod/requirements_byod_${PYTHON_VERSION}.in
- docker/base-deps/requirements.in
- docker/base-extra/requirements.in
constraints:
- /tmp/ray-deps/requirements_compiled_py${PYTHON_VERSION}.txt
# constraints:
# - /tmp/ray-deps/requirements_compiled_py${PYTHON_VERSION}.txt
depsets:
- ray_img_depset_${PYTHON_SHORT}
output: python/deplocks/base_extra_testdeps/ray-base_extra_testdeps_py${PYTHON_VERSION}.lock
Expand All @@ -67,6 +67,7 @@ depsets:
- --unsafe-package ray
- --python-version=${PYTHON_VERSION}
- --python-platform=linux
- --prerelease=allow
build_arg_sets:
- py311

Expand Down
1 change: 1 addition & 0 deletions ci/raydepsets/configs/rayllm.depsets.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@ build_arg_sets:
- --unsafe-package ray
- --python-platform=linux
- --extra-index-url https://download.pytorch.org/whl/${CUDA_CODE}
- --override python/requirements/llm/torch-override.txt
build_arg_sets:
- cpu
- cu128
Expand Down
4 changes: 2 additions & 2 deletions ci/raydepsets/pre_hooks/remove-compiled-headers.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,7 @@ fi

mkdir -p /tmp/ray-deps

# Remove the GPU constraints
# Remove the GPU constraints, numpy pin, and scipy pin (LLM requires numpy>=2 and compatible scipy)
cp "python/${FILENAME}" "/tmp/ray-deps/${FILENAME}"
sed -e '/^--extra-index-url /d' -e '/^--find-links /d' "/tmp/ray-deps/${FILENAME}" > "/tmp/ray-deps/${FILENAME}.tmp"
sed -e '/^--extra-index-url /d' -e '/^--find-links /d' -e '/^numpy==/d' -e '/^scipy==/d' "/tmp/ray-deps/${FILENAME}" > "/tmp/ray-deps/${FILENAME}.tmp"

@jeffreywang88 jeffreywang88 Jan 30, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

This is modified by Claude. We'll see if we need this.

mv "/tmp/ray-deps/${FILENAME}.tmp" "/tmp/ray-deps/${FILENAME}"
2 changes: 1 addition & 1 deletion doc/source/llm/doc_code/serve/multi_gpu/dp_basic_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,7 @@ def _testing_build_dp_openai_app(builder_config, **kwargs):
# Configure the model with data parallel settings
config = LLMConfig(
model_loading_config={
"model_id": "Qwen/Qwen2.5-0.5B-Instruct"
"model_id": "microsoft/Phi-tiny-MoE-instruct"
},
engine_kwargs={
"data_parallel_size": 2, # Number of DP replicas
Expand Down
4 changes: 2 additions & 2 deletions doc/source/llm/doc_code/serve/multi_gpu/dp_pd_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ def _testing_build_dp_deployment(llm_config, **kwargs):
# Configure prefill with data parallel attention
prefill_config = LLMConfig(
model_loading_config={
"model_id": "Qwen/Qwen2.5-0.5B-Instruct"
"model_id": "microsoft/Phi-tiny-MoE-instruct"
},
engine_kwargs={
"data_parallel_size": 2, # 2 DP replicas for prefill
Expand All @@ -78,7 +78,7 @@ def _testing_build_dp_deployment(llm_config, **kwargs):
# Configure decode with data parallel attention
decode_config = LLMConfig(
model_loading_config={
"model_id": "Qwen/Qwen2.5-0.5B-Instruct"
"model_id": "microsoft/Phi-tiny-MoE-instruct"
},
engine_kwargs={
"data_parallel_size": 2, # 2 DP replicas for decode (adjusted for 4 GPU limit)
Expand Down
2 changes: 1 addition & 1 deletion docker/ray-llm/Dockerfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ COPY python/deplocks/llm/rayllm_*.lock ./

# vLLM version tag to use for EP kernel and DeepGEMM install scripts
# Keep in sync with vllm version in python/requirements/llm/llm-requirements.txt
ARG VLLM_SCRIPTS_REF="v0.12.0"
ARG VLLM_SCRIPTS_REF="v0.14.0"

RUN <<EOF
#!/bin/bash
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_deps/ray_base_deps_py3.10.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_deps/ray_base_deps_py3.11.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_deps/ray_base_deps_py3.12.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_deps/ray_base_deps_py3.13.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_extra/ray_base_extra_py3.10.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_extra/ray_base_extra_py3.11.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_extra/ray_base_extra_py3.12.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
1 change: 1 addition & 0 deletions python/deplocks/base_extra/ray_base_extra_py3.13.lock
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,5 @@
--index-url https://pypi.org/simple
--extra-index-url http://localhost:9478/simple

adlfs==2023.8.0 \
--hash=sha256:07e804f6df4593acfcaf01025b162e30ac13e523d3570279c98b2d91a18026d9 \
Expand Down
Loading