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

[Refactor] Relocate completion and chat completion tests (#37125)

Browse Source 
Signed-off-by: sfeng33 <4florafeng@gmail.com>
This commit is contained in:
Flora Feng
2026-03-16 23:31:23 -04:00
committed by GitHub
parent f04d5226f8
commit 384dc7f77b
26 changed files with 41 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

150
tests/entrypoints/openai/chat_completion/test_vision_embeds.py Normal file
View File

@@ -0,0 +1,150 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import base64
import numpy as np
import pytest
import requests
import torch
from tests.utils import RemoteOpenAIServer
from vllm.utils.serial_utils import tensor2base64
@pytest.mark.parametrize(
"model_name", ["ibm-nasa-geospatial/Prithvi-EO-2.0-300M-TL-Sen1Floods11"]
)
def test_single_content(model_name: str):
args = [
"--runner",
"pooling",
# use half precision for speed and memory savings in CI environment
"--dtype",
"float16",
"--enforce-eager",
"--trust-remote-code",
"--max-num-seqs",
"32",
"--model-impl",
"terratorch",
"--skip-tokenizer-init",
"--enable-mm-embeds",
]
with RemoteOpenAIServer(model_name, args) as server:
response = requests.post(
server.url_for("pooling"),
json={
"model": model_name,
"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",
},
)
response.raise_for_status()
output = response.json()["data"][0]["data"]
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