Unverified Commit adf9bb3c authored by Sumanth R Hegde's avatar Sumanth R Hegde Committed by GitHub
Browse files

[CI] Add weight transfer tests to CI (#39821)


Signed-off-by: default avatarSumanthRH <sumanthrh99@gmail.com>
Co-authored-by: default avatarRobert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
Co-authored-by: default avatarCyrus Leung <tlleungac@connect.ust.hk>
parent b16fda62
...@@ -196,6 +196,8 @@ steps: ...@@ -196,6 +196,8 @@ steps:
- VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_async_new_apis.py - VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_async_new_apis.py
- VLLM_USE_DEEP_GEMM=1 VLLM_LOGGING_LEVEL=DEBUG python3 examples/offline_inference/data_parallel.py --model=Qwen/Qwen1.5-MoE-A2.7B -tp=1 -dp=2 --max-model-len=2048 --all2all-backend=deepep_high_throughput - VLLM_USE_DEEP_GEMM=1 VLLM_LOGGING_LEVEL=DEBUG python3 examples/offline_inference/data_parallel.py --model=Qwen/Qwen1.5-MoE-A2.7B -tp=1 -dp=2 --max-model-len=2048 --all2all-backend=deepep_high_throughput
- pytest -v -s tests/v1/distributed/test_dbo.py - pytest -v -s tests/v1/distributed/test_dbo.py
- VLLM_ALLOW_INSECURE_SERIALIZATION=1 pytest -v -s tests/distributed/test_weight_transfer.py
- pytest -v -s tests/distributed/test_packed_tensor.py
- label: Distributed Tests (2 GPUs)(B200) - label: Distributed Tests (2 GPUs)(B200)
device: b200 device: b200
......
...@@ -41,6 +41,7 @@ def create_mock_parallel_config( ...@@ -41,6 +41,7 @@ def create_mock_parallel_config(
config.rank = rank config.rank = rank
config.world_size = world_size config.world_size = world_size
config.data_parallel_rank = dp_rank config.data_parallel_rank = dp_rank
config.data_parallel_index = dp_rank
return config return config
...@@ -283,6 +284,7 @@ def inference_receive_tensor( ...@@ -283,6 +284,7 @@ def inference_receive_tensor(
parallel_config.rank = 0 parallel_config.rank = 0
parallel_config.world_size = 1 parallel_config.world_size = 1
parallel_config.data_parallel_rank = 0 parallel_config.data_parallel_rank = 0
parallel_config.data_parallel_index = 0
engine = NCCLWeightTransferEngine(config, parallel_config) engine = NCCLWeightTransferEngine(config, parallel_config)
...@@ -666,6 +668,7 @@ def inference_receive_ipc_tensor( ...@@ -666,6 +668,7 @@ def inference_receive_ipc_tensor(
parallel_config.rank = 0 parallel_config.rank = 0
parallel_config.world_size = 1 parallel_config.world_size = 1
parallel_config.data_parallel_rank = 0 parallel_config.data_parallel_rank = 0
parallel_config.data_parallel_index = 0
engine = IPCWeightTransferEngine(config, parallel_config) engine = IPCWeightTransferEngine(config, parallel_config)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment