mirror of
https://github.com/volcengine/verl.git
synced 2025-10-20 13:43:50 +08:00
[docs] refactor: use verl consistently in the codebase (#1390)
### Checklist Before Starting - [x] Search for similar PR(s). ### What does this PR do? Always use verl instead of veRL in the codebase, and add a CI check for this. ### Specific Changes mostly doc changes ### Test Added to sanity tests. ### Additional Info. - **Issue Number**: Fixes issue # or discussion # if any. - **Training**: [Note which backend this PR will affect: FSDP, Megatron, both, or none] - **Inference**: [Note which backend this PR will affect: vLLM, SGLang, both, or none] ### Checklist Before Submitting - [x] Read the [Contribute Guide](https://github.com/volcengine/verl?tab=readme-ov-file#contribution-guide). - [x] Apply [pre-commit checks](https://github.com/volcengine/verl?tab=readme-ov-file#code-linting-and-formatting). - [x] Add `[BREAKING]` to the PR title if it breaks any API. - [x] Update the documentation about your changes in the [docs](https://github.com/volcengine/verl/tree/main/docs). - [x] Add CI test(s) if neccessary. cc @ShaohonChen
This commit is contained in:
6
.github/workflows/sanity.yml
vendored
6
.github/workflows/sanity.yml
vendored
@ -46,3 +46,9 @@ jobs:
|
||||
- name: Run license test
|
||||
run: |
|
||||
python3 tests/sanity/check_license.py --directory .
|
||||
- name: Assert naming convention
|
||||
run: |
|
||||
if grep -rIn --exclude-dir=.git --exclude-dir=.github --exclude-dir=venv --exclude-dir=__pycache__ 'veRL' .; then
|
||||
echo "Please use verl instead of veRL in the codebase"
|
||||
exit 1
|
||||
fi
|
||||
|
@ -23,7 +23,7 @@ RUN pip3 install --no-cache-dir \
|
||||
|
||||
RUN pip3 install --no-cache-dir flash-attn==2.7.0.post2 --no-build-isolation
|
||||
|
||||
# vllm depends on ray, and verl does not support ray > 2.37
|
||||
# vllm depends on ray
|
||||
RUN pip3 install --no-cache-dir vllm==0.6.3 ray==2.10
|
||||
|
||||
# install apex
|
||||
|
@ -1,8 +1,10 @@
|
||||
# Upgrading to vllm >= 0.7
|
||||
|
||||
Note: verl+vllm 0.8.3 is now stable. Please see ``docs/README_vllm0.8.md`` for upgrade guide.
|
||||
|
||||
## Installation
|
||||
|
||||
Note: This version of verl+vllm 0.7+ supports **FSDP** for training and **vLLM** for rollout.
|
||||
Note: At time of writing, verl+vllm 0.7.x supports **FSDP** for training and **vLLM** for rollout.
|
||||
|
||||
```
|
||||
# Create the conda environment
|
||||
@ -68,4 +70,4 @@ VLLM_USE_PRECOMPILED=1 pip install --editable .
|
||||
```
|
||||
|
||||
Then you can enable the V1 engine by setting `export VLLM_USE_V1=1`. In some benchmark tests, the V1 engine demonstrates a 1.5x speed improvement over the vLLM V0 engine.
|
||||
The stable support of the vLLM V1 engine will come soon.
|
||||
The stable support of the vLLM V1 engine is available on verl main.
|
||||
|
@ -263,7 +263,7 @@ class ActorRolloutRefWorker(MegatronWorker):
|
||||
elif self.config.rollout.name == 'sglang':
|
||||
from verl.workers.rollout.sglang_rollout import SGLangRollout
|
||||
# NOTE(linjunrong): Due to recent fp8 support in SGLang. Now importing any symbol relate to SGLang's model_runner would check CUDA device capability.
|
||||
# However, due to veRL's setting, the main process of ray can not find any CUDA device, which would potentially lead to:
|
||||
# However, due to verl's setting, the main process of ray can not find any CUDA device, which would potentially lead to:
|
||||
# "RuntimeError: No CUDA GPUs are available".
|
||||
# For this reason, sharding_manager.__init__ should not import FSDPSGLangShardingManager and we import it here use the abs path.
|
||||
# check: https://github.com/sgl-project/sglang/blob/00f42707eaddfc2c0528e5b1e0094025c640b7a0/python/sglang/srt/layers/quantization/fp8_utils.py#L76
|
||||
|
Reference in New Issue
Block a user