@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / bcc6a09b63
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 16, 22:00 UTC

[CI/Build] Temporarily Remove Phi3-Vision from TP Test (#5989)

Roger Wang committed Jun 29, 2024 at 18:18 UTC bcc6a09b63aeb3efce964b54a756d431e580aebc
1 file changed +3 -2
.buildkite/test-pipeline.yaml
+3 -2
@@ -60,7 +60,8 @@ steps:
60 - TEST_DIST_MODEL=facebook/opt-125m DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_chunked_prefill_distributed.py
61 - TEST_DIST_MODEL=meta-llama/Llama-2-7b-hf DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_chunked_prefill_distributed.py
62 - TEST_DIST_MODEL=llava-hf/llava-1.5-7b-hf DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_multimodal_broadcast.py
63 - - TEST_DIST_MODEL=microsoft/Phi-3-vision-128k-instruct DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_multimodal_broadcast.py
63 + # FIXIT: find out why TP is failing with mp backend on phi3-v
64 + # - TEST_DIST_MODEL=microsoft/Phi-3-vision-128k-instruct DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_multimodal_broadcast.py
65 - pytest -v -s spec_decode/e2e/test_integration_dist_tp2.py
66 - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s test_sharded_state_loader.py
67 - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s distributed/test_utils.py
@@ -234,4 +235,4 @@ steps:
235 - TEST_DIST_MODEL=facebook/opt-125m DISTRIBUTED_EXECUTOR_BACKEND=mp pytest -v -s distributed/test_basic_distributed_correctness.py
236 - pip install https://github.com/flashinfer-ai/flashinfer/releases/download/v0.0.5/flashinfer-0.0.5+cu121torch2.3-cp310-cp310-linux_x86_64.whl
237 - VLLM_ATTENTION_BACKEND=FLASHINFER TEST_DIST_MODEL=facebook/opt-125m DISTRIBUTED_EXECUTOR_BACKEND=ray pytest -v -s distributed/test_basic_distributed_correctness.py
237 - - VLLM_ATTENTION_BACKEND=FLASHINFER TEST_DIST_MODEL=meta-llama/Meta-Llama-3-8B DISTRIBUTED_EXECUTOR_BACKEND=ray pytest -v -s distributed/test_basic_distributed_correctness.py
\ No newline at end of file
238 + - VLLM_ATTENTION_BACKEND=FLASHINFER TEST_DIST_MODEL=meta-llama/Meta-Llama-3-8B DISTRIBUTED_EXECUTOR_BACKEND=ray pytest -v -s distributed/test_basic_distributed_correctness.py