biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Convert formatting to use ruff instead of yapf + isort (#26247)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-10-05 15:06:22 +01:00
committed by GitHub
parent 17edd8a807
commit d6953beb91
1508 changed files with 115244 additions and 94146 deletions
Download Patch File Download Diff File Expand all files Collapse all files

117
tests/models/language/pooling/test_mm_classifier_conversion.py
View File

@@ -21,16 +21,18 @@ def test_idefics_multimodal(
"The future of AI is",
]
with vllm_runner(model_name="HuggingFaceM4/Idefics3-8B-Llama3",
runner="pooling",
task="classify",
convert="classify",
load_format="dummy",
max_model_len=512,
enforce_eager=True,
tensor_parallel_size=1,
disable_log_stats=True,
dtype="bfloat16") as vllm_model:
with vllm_runner(
model_name="HuggingFaceM4/Idefics3-8B-Llama3",
runner="pooling",
task="classify",
convert="classify",
load_format="dummy",
max_model_len=512,
enforce_eager=True,
tensor_parallel_size=1,
disable_log_stats=True,
dtype="bfloat16",
) as vllm_model:
llm = vllm_model.get_llm()
outputs = llm.classify(prompts)
for output in outputs:
@@ -38,19 +40,20 @@ def test_idefics_multimodal(
def update_config(config):
config.text_config.update({
"architectures": ["Gemma3ForSequenceClassification"],
"classifier_from_token": ["A", "B", "C", "D", "E"],
"method":
"no_post_processing",
"id2label": {
"A": "Chair",
"B": "Couch",
"C": "Table",
"D": "Bed",
"E": "Cupboard"
},
})
config.text_config.update(
{
"architectures": ["Gemma3ForSequenceClassification"],
"classifier_from_token": ["A", "B", "C", "D", "E"],
"method": "no_post_processing",
"id2label": {
"A": "Chair",
"B": "Couch",
"C": "Table",
"D": "Bed",
"E": "Cupboard",
},
}
)
return config
@@ -63,11 +66,10 @@ def test_gemma_multimodal(
# switch to use ROCm CK FA backend
monkeypatch.setenv("VLLM_USE_TRITON_FLASH_ATTN", "False")
messages = [{
"role":
"system",
"content":
"""
messages = [
{
"role": "system",
"content": """
You are a helpful assistant. You will be given a product description
which may also include an image. Classify the following product into
one of the categories:
@@ -78,38 +80,39 @@ def test_gemma_multimodal(
D = bed
E = cupboard
You'll answer with exactly one letter (A, B, C, D, or E)."""
}, {
"role":
"user",
"content": [{
"type": "image_url",
"image_url": {
"url":
"https://upload.wikimedia.org/wikipedia/commons/c/c6/Set_of_fourteen_side_chairs_MET_DP110780.jpg"
}
}, {
"type": "text",
"text": "A fine 19th century piece of furniture."
}]
}]
with vllm_runner(model_name="google/gemma-3-4b-it",
runner="pooling",
task="classify",
convert="classify",
load_format="auto",
hf_overrides=update_config,
pooler_config=PoolerConfig(pooling_type="LAST"),
max_model_len=512,
enforce_eager=True,
tensor_parallel_size=1,
disable_log_stats=True,
dtype="bfloat16") as vllm_model:
You'll answer with exactly one letter (A, B, C, D, or E).""",
},
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {
"url": "https://upload.wikimedia.org/wikipedia/commons/c/c6/Set_of_fourteen_side_chairs_MET_DP110780.jpg"
},
},
{"type": "text", "text": "A fine 19th century piece of furniture."},
],
},
]
with vllm_runner(
model_name="google/gemma-3-4b-it",
runner="pooling",
task="classify",
convert="classify",
load_format="auto",
hf_overrides=update_config,
pooler_config=PoolerConfig(pooling_type="LAST"),
max_model_len=512,
enforce_eager=True,
tensor_parallel_size=1,
disable_log_stats=True,
dtype="bfloat16",
) as vllm_model:
llm = vllm_model.get_llm()
prompts = llm.preprocess_chat(messages)
result = llm.classify(prompts)
assert result[0].outputs.probs[0] > 0.95
assert all(c < 0.05 for c in result[0].outputs.probs[1:])
assert all(c < 0.05 for c in result[0].outputs.probs[1:])