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

[Refactor] Use data parser for matching data items to multi-modal UUIDs (#32955)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-01-26 15:00:28 +08:00
committed by GitHub
parent ee484b3f4b
commit 11b556878b
14 changed files with 701 additions and 604 deletions
Download Patch File Download Diff File Expand all files Collapse all files

126
tests/entrypoints/openai/test_vision_embeds.py
View File

@@ -13,31 +13,10 @@ from vllm.utils.serial_utils import tensor2base64
from ...utils import RemoteOpenAIServer
def _terratorch_dummy_messages():
pixel_values = torch.full((6, 512, 512), 1.0, dtype=torch.float16)
location_coords = torch.full((1, 2), 1.0, dtype=torch.float16)
return [
{
"role": "user",
"content": [
{
"type": "image_embeds",
"image_embeds": {
"pixel_values": tensor2base64(pixel_values),
"location_coords": tensor2base64(location_coords),
},
}
],
}
]
@pytest.mark.asyncio
@pytest.mark.parametrize(
"model_name", ["ibm-nasa-geospatial/Prithvi-EO-2.0-300M-TL-Sen1Floods11"]
)
def test_single_request(model_name: str):
def test_single_content(model_name: str):
args = [
"--runner",
"pooling",
@@ -59,7 +38,24 @@ def test_single_request(model_name: str):
server.url_for("pooling"),
json={
"model": model_name,
"messages": _terratorch_dummy_messages(),
"messages": [
{
"role": "user",
"content": [
{
"type": "image_embeds",
"image_embeds": {
"pixel_values": tensor2base64(
torch.ones((6, 512, 512), dtype=torch.float16)
),
"location_coords": tensor2base64(
torch.ones((1, 2), dtype=torch.float16)
),
},
},
],
}
],
"encoding_format": "base64",
},
)
@@ -69,3 +65,87 @@ def test_single_request(model_name: str):
np_response = np.frombuffer(base64.b64decode(output), dtype=np.float32)
assert len(np_response) == 524288
@pytest.mark.parametrize("model_name", ["Qwen/Qwen3-VL-2B-Instruct"])
def test_multi_content(model_name: str):
args = [
"--enforce-eager",
"--max-num-seqs",
"32",
"--max-model-len",
"8192",
"--enable-mm-embeds",
]
with RemoteOpenAIServer(model_name, args) as server:
client = server.get_client()
# Image only
chat_completion = client.chat.completions.create(
model=model_name,
messages=[
{
"role": "user",
"content": [
{
"type": "image_embeds",
"image_embeds": {
"image_embeds": tensor2base64(torch.zeros(220, 8192)),
"image_grid_thw": tensor2base64(
torch.tensor([1, 22, 40])
),
},
},
{
"type": "image_embeds",
"image_embeds": {
"image_embeds": tensor2base64(torch.zeros(220, 8192)),
"image_grid_thw": tensor2base64(
torch.tensor([1, 22, 40])
),
},
},
],
}
],
max_tokens=5,
)
assert chat_completion.id is not None
assert len(chat_completion.choices) == 1
# Interleaved text and image
chat_completion = client.chat.completions.create(
model=model_name,
messages=[
{
"role": "user",
"content": [
{
"type": "image_embeds",
"image_embeds": {
"image_embeds": tensor2base64(torch.zeros(220, 8192)),
"image_grid_thw": tensor2base64(
torch.tensor([1, 22, 40])
),
},
},
{"type": "text", "text": "OCR:"},
{
"type": "image_embeds",
"image_embeds": {
"image_embeds": tensor2base64(torch.zeros(220, 8192)),
"image_grid_thw": tensor2base64(
torch.tensor([1, 22, 40])
),
},
},
],
}
],
max_tokens=5,
)
assert chat_completion.id is not None
assert len(chat_completion.choices) == 1