Update handler.py
Browse files- handler.py +12 -45
handler.py
CHANGED
@@ -1,40 +1,18 @@
|
|
1 |
from typing import Dict, Any
|
2 |
import torch
|
3 |
-
from transformers import
|
4 |
from PIL import Image
|
5 |
import io
|
6 |
import base64
|
7 |
import requests
|
8 |
-
from qwen_vl_utils import process_vision_info
|
9 |
|
10 |
class EndpointHandler():
|
11 |
def __init__(self, path=""):
|
12 |
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
13 |
-
self.model = Qwen2VLForConditionalGeneration.from_pretrained(
|
14 |
-
path,
|
15 |
-
torch_dtype="auto",
|
16 |
-
device_map="auto"
|
17 |
-
).to(self.device)
|
18 |
-
|
19 |
self.processor = AutoProcessor.from_pretrained(path)
|
20 |
-
|
21 |
-
# Optionally, adjust min_pixels and max_pixels if needed
|
22 |
-
# min_pixels = 256*28*28
|
23 |
-
# max_pixels = 1280*28*28
|
24 |
-
# self.processor = AutoProcessor.from_pretrained(path, min_pixels=min_pixels, max_pixels=max_pixels)
|
25 |
|
26 |
def __call__(self, data: Any) -> Dict[str, Any]:
|
27 |
-
"""
|
28 |
-
Args:
|
29 |
-
data (Any): The input data, which can be:
|
30 |
-
- Binary image data in the request body.
|
31 |
-
- A dictionary with 'image' and 'text' keys:
|
32 |
-
- 'image': Base64-encoded image string or image URL.
|
33 |
-
- 'text': The text prompt.
|
34 |
-
|
35 |
-
Returns:
|
36 |
-
Dict[str, Any]: The generated text output from the model.
|
37 |
-
"""
|
38 |
default_prompt = "Describe this image."
|
39 |
|
40 |
if isinstance(data, (bytes, bytearray)):
|
@@ -46,8 +24,7 @@ class EndpointHandler():
|
|
46 |
if image_input is None:
|
47 |
return {"error": "No image provided."}
|
48 |
if image_input.startswith('http'):
|
49 |
-
|
50 |
-
image = Image.open(io.BytesIO(response.content)).convert('RGB')
|
51 |
else:
|
52 |
image_data = base64.b64decode(image_input)
|
53 |
image = Image.open(io.BytesIO(image_data)).convert('RGB')
|
@@ -58,34 +35,24 @@ class EndpointHandler():
|
|
58 |
{
|
59 |
"role": "user",
|
60 |
"content": [
|
61 |
-
{
|
62 |
-
"type": "image",
|
63 |
-
"image": image,
|
64 |
-
},
|
65 |
{"type": "text", "text": text_input},
|
66 |
],
|
67 |
}
|
68 |
]
|
69 |
|
70 |
-
text = self.processor.apply_chat_template(
|
71 |
-
messages, tokenize=False, add_generation_prompt=True
|
72 |
-
)
|
73 |
-
image_inputs, video_inputs = process_vision_info(messages)
|
74 |
inputs = self.processor(
|
75 |
text=[text],
|
76 |
-
images=
|
77 |
-
videos=video_inputs,
|
78 |
padding=True,
|
79 |
return_tensors="pt",
|
80 |
-
)
|
81 |
-
inputs = inputs.to(self.device)
|
82 |
|
83 |
-
|
84 |
-
generated_ids_trimmed = [
|
85 |
-
out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
|
86 |
-
]
|
87 |
output_text = self.processor.batch_decode(
|
88 |
-
|
89 |
-
)
|
90 |
|
91 |
-
return {"generated_text": output_text
|
|
|
|
1 |
from typing import Dict, Any
|
2 |
import torch
|
3 |
+
from transformers import AutoProcessor, Qwen2VLForConditionalGeneration
|
4 |
from PIL import Image
|
5 |
import io
|
6 |
import base64
|
7 |
import requests
|
|
|
8 |
|
9 |
class EndpointHandler():
|
10 |
def __init__(self, path=""):
|
11 |
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
12 |
+
self.model = Qwen2VLForConditionalGeneration.from_pretrained(path).to(self.device)
|
|
|
|
|
|
|
|
|
|
|
13 |
self.processor = AutoProcessor.from_pretrained(path)
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
def __call__(self, data: Any) -> Dict[str, Any]:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
default_prompt = "Describe this image."
|
17 |
|
18 |
if isinstance(data, (bytes, bytearray)):
|
|
|
24 |
if image_input is None:
|
25 |
return {"error": "No image provided."}
|
26 |
if image_input.startswith('http'):
|
27 |
+
image = Image.open(requests.get(image_input, stream=True).raw).convert('RGB')
|
|
|
28 |
else:
|
29 |
image_data = base64.b64decode(image_input)
|
30 |
image = Image.open(io.BytesIO(image_data)).convert('RGB')
|
|
|
35 |
{
|
36 |
"role": "user",
|
37 |
"content": [
|
38 |
+
{"type": "image", "image": image},
|
|
|
|
|
|
|
39 |
{"type": "text", "text": text_input},
|
40 |
],
|
41 |
}
|
42 |
]
|
43 |
|
44 |
+
text = self.processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
|
|
|
|
|
|
45 |
inputs = self.processor(
|
46 |
text=[text],
|
47 |
+
images=[image],
|
|
|
48 |
padding=True,
|
49 |
return_tensors="pt",
|
50 |
+
).to(self.device)
|
|
|
51 |
|
52 |
+
generate_ids = self.model.generate(inputs.input_ids, max_length=30)
|
|
|
|
|
|
|
53 |
output_text = self.processor.batch_decode(
|
54 |
+
generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False
|
55 |
+
)[0]
|
56 |
|
57 |
+
return {"generated_text": output_text}
|
58 |
+
|