Skip to content

feat: support non-colocated sync vllm #489

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
chtruong814 merged 8 commits into from
Jun 12, 2025
Merged

feat: support non-colocated sync vllm #489

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
43fd588
support non colocated in unit test with fake rank
yuki-97 Jun 10, 2025
10f5b3e
correct rank and add inference gpu settings
yuki-97 Jun 10, 2025
fe519c8
no need to offload when non-colocated
yuki-97 Jun 6, 2025
2c195bf
fix vllm tp-size>1 by using vllm collective
yuki-97 Jun 9, 2025
a803dbc
update config
yuki-97 Jun 10, 2025
2c7f765
update comment
yuki-97 Jun 12, 2025
baf3f29
dtensor worker use vllm
yuki-97 Jun 12, 2025
21b6193
add functional test
yuki-97 Jun 12, 2025
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
1 change: 1 addition & 0 deletions .github/workflows/cicd-main.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -188,6 +188,7 @@ jobs:
time uv run --no-sync bash ./tests/functional/sft.sh
time uv run --no-sync bash ./tests/functional/grpo.sh
time uv run --no-sync bash ./tests/functional/grpo_multiturn.sh
time uv run --no-sync bash ./tests/functional/grpo_non_colocated.sh
time uv run --no-sync bash ./tests/functional/dpo.sh
time uv run --no-sync bash ./tests/functional/eval.sh
time uv run --no-sync bash ./tests/functional/test_mcore_extra_installed_correctly.sh
Expand Down
8 changes: 8 additions & 0 deletions examples/configs/grpo-deepscaler-1.5b-8K.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,6 +101,14 @@ policy:
# For most cases, use "dummy" to load the initial weights, since they will be overwritten during refit
# For Gemma models, we need to use "auto" due to a vllm bug
load_format: dummy
colocated:
Comment thread
yuki-97 marked this conversation as resolved.
# true: generation shares training GPUs
# false: uses dedicated generation resources
enabled: true
# only relevant when enabled is false
resources:
gpus_per_node: null # Decides num gpus to be dedicated to generation when there is one node in the cluster i.e cluster.num_nodes == 1
num_nodes: null # Decides number of nodes to be dedicated to generation

data:
max_input_seq_length: ${policy.max_total_sequence_length} # upper bound, real truncation occurs at vllm.max_model_len
Expand Down
8 changes: 8 additions & 0 deletions examples/configs/grpo_math_1B.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -106,6 +106,14 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: ${policy.max_total_sequence_length}
colocated:
# true: generation shares training GPUs
# false: uses dedicated generation resources
enabled: true
# only relevant when enabled is false
resources:
gpus_per_node: null # Decides num gpus to be dedicated to generation when there is one node in the cluster i.e cluster.num_nodes == 1
num_nodes: null # Decides number of nodes to be dedicated to generation

data:
max_input_seq_length: ${policy.max_total_sequence_length} # upper bound, real truncation occurs at vllm.max_model_len
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-gemma3-1b-it-1n8g-fsdp2tp1.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,6 +88,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 512
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 512
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-gemma3-27b-it-16n8g-fsdp2tp8sp-actckpt-long.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 16384
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 16384
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-llama3.1-8b-instruct-4n8g-fsdp2tp1-long.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 4096
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 4096
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-llama3.2-1b-instruct-1n8g-fsdp2tp1.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 512
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 512
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-qwen2.5-32b-16n8g-fsdp2tp8sp-actckpt-long.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 16384
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 16384
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-qwen2.5-32b-16n8g-fsdp2tp8sp-actckpt.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 16384
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 16384
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-qwen2.5-7b-instruct-4n8g-fsdp1.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,6 +86,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 4096
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 4096
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-qwen2.5-7b-instruct-4n8g-fsdp2tp4sp.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 4096
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 4096
prompt_file: examples/prompts/cot.txt
Expand Down
5 changes: 5 additions & 0 deletions examples/configs/recipes/llm/grpo-qwen2.5-math-1.5b-instruct-1n8g-fsdp2tp1.v3.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,11 @@ policy:
pipeline_parallel_size: 1
gpu_memory_utilization: 0.6
max_model_len: 512
colocated:
enabled: true
resources:
gpus_per_node: null
num_nodes: null
data:
max_input_seq_length: 512
prompt_file: examples/prompts/cot.txt
Expand Down
Loading
Loading