Compare commits

...

3 Commits

18 changed files with 34 additions and 13 deletions

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/dataset.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/dataset.yml
@ -35,4 +37,4 @@ jobs:
- name: Running ray test using cupy (move it to L20 when dockerfile ready)
run: |
cd tests/ray
pytest -s -x test_rvdz.py
pytest -s -x test_rvdz.py

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_digit_completion.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_digit_completion.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_gsm8k.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_gsm8k.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_gsm8k_megatron.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_gsm8k_megatron.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_lora.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_lora.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_sft.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/e2e_sft.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/model.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/model.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/ray_test.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/ray_test.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/sandbox.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/sandbox.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/sanity.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/sanity.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/vllm.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/vllm.yml

View File

@ -6,12 +6,14 @@ on:
push:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/yapf_format.yml
pull_request:
branches:
- main
- v0.2.x
paths:
- "**/*.py"
- .github/workflows/yapf_format.yml

View File

@ -38,9 +38,8 @@ Image and tag: ``verlai/verl:vemlp-th2.4.0-cu124-vllm0.6.3-ray2.10-te1.7-v0.0.3`
.. code:: bash
# install the nightly version (recommended)
git clone https://github.com/volcengine/verl && cd verl && pip3 install -e .
# or install from pypi via `pip3 install verl`
# install the stable version
pip3 install verl
3. Setup Megatron (optional)
@ -83,9 +82,7 @@ own post-training jobs.
# install verl together with some lightweight dependencies in setup.py
pip3 install torch==2.4.0 --index-url https://download.pytorch.org/whl/cu124
pip3 install flash-attn --no-build-isolation
git clone https://github.com/volcengine/verl.git
cd verl
pip3 install -e .
pip3 install verl
Megatron is optional. It's dependencies can be setup as below:

View File

@ -9,7 +9,7 @@ python3 -m verl.trainer.main_ppo \
data.max_response_length=512 \
actor_rollout_ref.model.path=google/gemma-2-2b-it \
actor_rollout_ref.actor.optim.lr=1e-6 \
actor_rollout_ref.model.use_remove_padding=True \
actor_rollout_ref.model.use_remove_padding=False \
actor_rollout_ref.actor.ppo_mini_batch_size=128 \
actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \
actor_rollout_ref.actor.fsdp_config.param_offload=False \
@ -22,7 +22,7 @@ python3 -m verl.trainer.main_ppo \
actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4 \
actor_rollout_ref.ref.fsdp_config.param_offload=True \
critic.optim.lr=1e-5 \
critic.model.use_remove_padding=True \
critic.model.use_remove_padding=False \
critic.model.path=google/gemma-2-2b-it \
critic.model.enable_gradient_checkpointing=False \
critic.ppo_micro_batch_size_per_gpu=4 \

View File

@ -44,7 +44,7 @@ dependencies = [
"ray>=2.10",
"tensordict<0.6",
"transformers",
"vllm<=0.6.3.post1",
"vllm<=0.6.3",
'wandb',
]

View File

@ -15,5 +15,5 @@ pylatexenc
ray
tensordict<0.6
transformers
vllm==0.6.3.post1
vllm<=0.6.3
wandb

View File

@ -36,7 +36,7 @@ install_requires = [
'ray>=2.10',
'tensordict<0.6',
'transformers',
'vllm<=0.6.3.post1',
'vllm<=0.6.3',
'wandb',
]

View File

@ -1 +1 @@
0.2
0.2.0.post2