mirror of
https://git.datalinker.icu/vllm-project/vllm.git
synced 2025-12-13 19:15:34 +08:00
8a7cc254a0
Lora 3 & 4 test seems to have illegal memory access failure after this commit; [2024-05-14 23:51:18,182 E 22 22] logging.cc:101: Unhandled exception: N3c105ErrorE. what(): CUDA error: an illegal memory access was encountered <br class="Apple-interchange-newline"> Exmaple: https://buildkite.com/vllm/ci/builds/7382#018f793d-1527-4e1c-ab59-c3a34ec55241 This reverts commit 1356df5. FILL IN THE PR DESCRIPTION HERE FIX #xxxx (link existing issues this PR will resolve)
61 lines
1.8 KiB
Python
61 lines
1.8 KiB
Python
"""Compare the outputs of HF and vLLM when using greedy sampling.
|
|
|
|
This tests bigger models and use half precision.
|
|
|
|
Run `pytest tests/models/test_big_models.py`.
|
|
"""
|
|
import pytest
|
|
|
|
MODELS = [
|
|
"meta-llama/Llama-2-7b-hf",
|
|
# "mistralai/Mistral-7B-v0.1", # Tested by test_mistral.py
|
|
# "Deci/DeciLM-7b", # Broken
|
|
# "tiiuae/falcon-7b", # Broken
|
|
"EleutherAI/gpt-j-6b",
|
|
"mosaicml/mpt-7b",
|
|
# "Qwen/Qwen1.5-0.5B" # Broken,
|
|
]
|
|
|
|
|
|
@pytest.mark.parametrize("model", MODELS)
|
|
@pytest.mark.parametrize("dtype", ["half"])
|
|
@pytest.mark.parametrize("max_tokens", [32])
|
|
def test_models(
|
|
hf_runner,
|
|
vllm_runner,
|
|
example_prompts,
|
|
model: str,
|
|
dtype: str,
|
|
max_tokens: int,
|
|
) -> None:
|
|
hf_model = hf_runner(model, dtype=dtype)
|
|
hf_outputs = hf_model.generate_greedy(example_prompts, max_tokens)
|
|
del hf_model
|
|
|
|
vllm_model = vllm_runner(model, dtype=dtype)
|
|
vllm_outputs = vllm_model.generate_greedy(example_prompts, max_tokens)
|
|
del vllm_model
|
|
|
|
for i in range(len(example_prompts)):
|
|
hf_output_ids, hf_output_str = hf_outputs[i]
|
|
vllm_output_ids, vllm_output_str = vllm_outputs[i]
|
|
assert hf_output_str == vllm_output_str, (
|
|
f"Test{i}:\nHF: {hf_output_str!r}\nvLLM: {vllm_output_str!r}")
|
|
assert hf_output_ids == vllm_output_ids, (
|
|
f"Test{i}:\nHF: {hf_output_ids}\nvLLM: {vllm_output_ids}")
|
|
|
|
|
|
@pytest.mark.parametrize("model", MODELS)
|
|
@pytest.mark.parametrize("dtype", ["half"])
|
|
def test_model_print(
|
|
vllm_runner,
|
|
model: str,
|
|
dtype: str,
|
|
) -> None:
|
|
vllm_model = vllm_runner(model, dtype=dtype)
|
|
# This test is for verifying whether the model's extra_repr
|
|
# can be printed correctly.
|
|
print(vllm_model.model.llm_engine.model_executor.driver_worker.
|
|
model_runner.model)
|
|
del vllm_model
|