vllm/tests/models/embedding/language/test_embedding.py

52 lines
1.8 KiB
Python

"""Compare the outputs of HF and vLLM for Mistral models using greedy sampling.
Run `pytest tests/models/embedding/language/test_embedding.py`.
"""
import pytest
import torch
import torch.nn.functional as F
MODELS = [
"intfloat/e5-mistral-7b-instruct",
"BAAI/bge-multilingual-gemma2",
]
def compare_embeddings(embeddings1, embeddings2):
similarities = [
F.cosine_similarity(torch.tensor(e1), torch.tensor(e2), dim=0)
for e1, e2 in zip(embeddings1, embeddings2)
]
return similarities
@pytest.mark.parametrize("model", MODELS)
@pytest.mark.parametrize("dtype", ["half"])
def test_models(
hf_runner,
vllm_runner,
example_prompts,
model: str,
dtype: str,
) -> None:
# The example_prompts has ending "\n", for example:
# "Write a short story about a robot that dreams for the first time.\n"
# sentence_transformers will strip the input texts, see:
# https://github.com/UKPLab/sentence-transformers/blob/v3.1.1/sentence_transformers/models/Transformer.py#L159
# This makes the input_ids different between hf_model and vllm_model.
# So we need to strip the input texts to avoid test failing.
example_prompts = [str(s).strip() for s in example_prompts]
with hf_runner(model, dtype=dtype, is_embedding_model=True) as hf_model:
hf_outputs = hf_model.encode(example_prompts)
with vllm_runner(model, dtype=dtype) as vllm_model:
vllm_outputs = vllm_model.encode(example_prompts)
similarities = compare_embeddings(hf_outputs, vllm_outputs)
all_similarities = torch.stack(similarities)
tolerance = 1e-2
assert torch.all((all_similarities <= 1.0 + tolerance)
& (all_similarities >= 1.0 - tolerance)
), f"Not all values are within {tolerance} of 1.0"