Skip to content

[ci][distributed] fix phi-3v test failure #5990

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
wants to merge 3 commits into from
Closed

[ci][distributed] fix phi-3v test failure #5990

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
cd8d0f5
temp fix test bug
youkaichao Jun 30, 2024
54331fc
restore
youkaichao Jun 30, 2024
01d2165
adjust run order with hf and vllm
youkaichao Jun 30, 2024
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
5 changes: 4 additions & 1 deletion .buildkite/test-pipeline.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,7 +60,10 @@ steps:
- TEST_DIST_MODEL=facebook/opt-125m DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_chunked_prefill_distributed.py
- TEST_DIST_MODEL=meta-llama/Llama-2-7b-hf DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_chunked_prefill_distributed.py
- TEST_DIST_MODEL=llava-hf/llava-1.5-7b-hf DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_multimodal_broadcast.py
- TEST_DIST_MODEL=microsoft/Phi-3-vision-128k-instruct DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_multimodal_broadcast.py
# FIXIT: the phi-3v test requires using fork method
# see https://buildkite.com/vllm/ci-aws/builds/3286#019064f5-697e-4477-ad66-b97b2763e78c for reference
# need to investigate and control when we import the huggingface config class
- VLLM_WORKER_MULTIPROC_METHOD=fork TEST_DIST_MODEL=microsoft/Phi-3-vision-128k-instruct DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_multimodal_broadcast.py
- pytest -v -s spec_decode/e2e/test_integration_dist_tp2.py
- CUDA_VISIBLE_DEVICES=0,1 pytest -v -s test_sharded_state_loader.py
- CUDA_VISIBLE_DEVICES=0,1 pytest -v -s distributed/test_utils.py
Expand Down
22 changes: 12 additions & 10 deletions tests/models/test_phi3v.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,16 +97,8 @@ def run_test(
hf_images = [asset.for_hf() for asset in image_assets]
vllm_images = [asset.for_vllm(vlm_config) for asset in image_assets]

# use eager mode for hf runner, since phi3_v didn't work with flash_attn
hf_model_kwargs = {"_attn_implementation": "eager"}
with hf_runner(model_id, dtype=dtype,
model_kwargs=hf_model_kwargs) as hf_model:
hf_outputs = hf_model.generate_greedy(
HF_IMAGE_PROMPTS,
max_tokens,
images=hf_images,
eos_token_id=hf_model.processor.tokenizer.eos_token_id)

# NOTE: run vLLM first, and then run HF
# otherwise, HF will initialize the GPU and vLLM will fail with fork
vllm_image_prompts = [
p.replace("<|image_1|>",
"<|image|>" * vlm_config.image_feature_size + "<s>")
Expand All @@ -124,6 +116,16 @@ def run_test(
max_tokens,
images=vllm_images)

# use eager mode for hf runner, since phi3_v didn't work with flash_attn
hf_model_kwargs = {"_attn_implementation": "eager"}
with hf_runner(model_id, dtype=dtype,
model_kwargs=hf_model_kwargs) as hf_model:
hf_outputs = hf_model.generate_greedy(
HF_IMAGE_PROMPTS,
max_tokens,
images=hf_images,
eos_token_id=hf_model.processor.tokenizer.eos_token_id)

for i in range(len(HF_IMAGE_PROMPTS)):
hf_output_ids, hf_output_str = hf_outputs[i]
vllm_output_ids, vllm_output_str = vllm_to_hf_output(
Expand Down