forked from sgl-project/sglang
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_vision_openai_server.py
290 lines (256 loc) · 9.79 KB
/
test_vision_openai_server.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
import base64
import io
import json
import os
import unittest
import numpy as np
import openai
import requests
from decord import VideoReader, cpu
from PIL import Image
from sglang.srt.utils import kill_child_process
from sglang.test.test_utils import (
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
DEFAULT_URL_FOR_TEST,
popen_launch_server,
)
class TestOpenAIVisionServer(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.model = "lmms-lab/llava-onevision-qwen2-0.5b-ov"
cls.base_url = DEFAULT_URL_FOR_TEST
cls.api_key = "sk-123456"
cls.process = popen_launch_server(
cls.model,
cls.base_url,
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
api_key=cls.api_key,
other_args=[
"--chat-template",
"chatml-llava",
# "--log-requests",
],
)
cls.base_url += "/v1"
@classmethod
def tearDownClass(cls):
kill_child_process(cls.process.pid)
def test_chat_completion(self):
client = openai.Client(api_key=self.api_key, base_url=self.base_url)
response = client.chat.completions.create(
model="default",
messages=[
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {
"url": "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
},
},
{
"type": "text",
"text": "Describe this image in a very short sentence.",
},
],
},
],
temperature=0,
)
assert response.choices[0].message.role == "assistant"
text = response.choices[0].message.content
assert isinstance(text, str)
assert "man" in text or "cab" in text, text
assert response.id
assert response.created
assert response.usage.prompt_tokens > 0
assert response.usage.completion_tokens > 0
assert response.usage.total_tokens > 0
def test_multi_turn_chat_completion(self):
client = openai.Client(api_key=self.api_key, base_url=self.base_url)
response = client.chat.completions.create(
model="default",
messages=[
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {
"url": "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
},
},
{
"type": "text",
"text": "Describe this image in a very short sentence.",
},
],
},
{
"role": "assistant",
"content": [
{
"type": "text",
"text": "There is a man at the back of a yellow cab ironing his clothes.",
}
],
},
{
"role": "user",
"content": [
{"type": "text", "text": "Repeat your previous answer."}
],
},
],
temperature=0,
)
assert response.choices[0].message.role == "assistant"
text = response.choices[0].message.content
assert isinstance(text, str)
assert "man" in text or "cab" in text, text
assert response.id
assert response.created
assert response.usage.prompt_tokens > 0
assert response.usage.completion_tokens > 0
assert response.usage.total_tokens > 0
def test_mult_images_chat_completion(self):
client = openai.Client(api_key=self.api_key, base_url=self.base_url)
response = client.chat.completions.create(
model="default",
messages=[
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {
"url": "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png"
},
},
{
"type": "image_url",
"image_url": {
"url": "https://raw.githubusercontent.com/sgl-project/sglang/main/assets/logo.png"
},
},
{
"type": "text",
"text": "I have two very different images. They are not related at all. "
"Please describe the first image in one sentence, and then describe the second image in another sentence.",
},
],
},
],
temperature=0,
)
assert response.choices[0].message.role == "assistant"
text = response.choices[0].message.content
assert isinstance(text, str)
print(text)
assert "man" in text and "taxi" in text, text
assert "logo" in text, text
assert response.id
assert response.created
assert response.usage.prompt_tokens > 0
assert response.usage.completion_tokens > 0
assert response.usage.total_tokens > 0
def prepare_video_messages(self, video_path):
max_frames_num = 32
vr = VideoReader(video_path, ctx=cpu(0))
total_frame_num = len(vr)
uniform_sampled_frames = np.linspace(
0, total_frame_num - 1, max_frames_num, dtype=int
)
frame_idx = uniform_sampled_frames.tolist()
frames = vr.get_batch(frame_idx).asnumpy()
base64_frames = []
for frame in frames:
pil_img = Image.fromarray(frame)
buff = io.BytesIO()
pil_img.save(buff, format="JPEG")
base64_str = base64.b64encode(buff.getvalue()).decode("utf-8")
base64_frames.append(base64_str)
messages = [{"role": "user", "content": []}]
frame_format = {
"type": "image_url",
"image_url": {"url": "data:image/jpeg;base64,{}"},
}
for base64_frame in base64_frames:
frame_format["image_url"]["url"] = "data:image/jpeg;base64,{}".format(
base64_frame
)
messages[0]["content"].append(frame_format.copy())
prompt = {"type": "text", "text": "Please describe the video in detail."}
messages[0]["content"].append(prompt)
return messages
def test_video_chat_completion(self):
url = "https://raw.githubusercontent.com/EvolvingLMMs-Lab/sglang/dev/onevision_local/assets/jobs.mp4"
cache_dir = os.path.expanduser("~/.cache")
file_path = os.path.join(cache_dir, "jobs.mp4")
os.makedirs(cache_dir, exist_ok=True)
if not os.path.exists(file_path):
response = requests.get(url)
response.raise_for_status()
with open(file_path, "wb") as f:
f.write(response.content)
client = openai.Client(api_key=self.api_key, base_url=self.base_url)
messages = self.prepare_video_messages(file_path)
video_request = client.chat.completions.create(
model="default",
messages=messages,
temperature=0,
max_tokens=1024,
stream=True,
)
print("-" * 30)
video_response = ""
for chunk in video_request:
if chunk.choices[0].delta.content is not None:
content = chunk.choices[0].delta.content
video_response += content
print(content, end="", flush=True)
print("-" * 30)
# Add assertions to validate the video response
self.assertIsNotNone(video_response)
self.assertGreater(len(video_response), 0)
def test_regex(self):
client = openai.Client(api_key=self.api_key, base_url=self.base_url)
regex = (
r"""\{\n"""
+ r""" "color": "[\w]+",\n"""
+ r""" "number_of_cars": [\d]+\n"""
+ r"""\}"""
)
response = client.chat.completions.create(
model="default",
messages=[
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {
"url": "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
},
},
{
"type": "text",
"text": "Describe this image in the JSON format.",
},
],
},
],
temperature=0,
extra_body={"regex": regex},
)
text = response.choices[0].message.content
try:
js_obj = json.loads(text)
except (TypeError, json.decoder.JSONDecodeError):
print("JSONDecodeError", text)
raise
assert isinstance(js_obj["color"], str)
assert isinstance(js_obj["number_of_cars"], int)
if __name__ == "__main__":
unittest.main()