Update app.py
Browse files
app.py
CHANGED
@@ -1,7 +1,8 @@
|
|
1 |
import os
|
2 |
-
from
|
|
|
3 |
from pydantic import BaseModel
|
4 |
-
from typing import
|
5 |
import torch
|
6 |
from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
|
7 |
from qwen_vl_utils import process_vision_info
|
@@ -9,50 +10,66 @@ from byaldi import RAGMultiModalModel
|
|
9 |
from PIL import Image
|
10 |
import io
|
11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
# Initialize FastAPI app
|
13 |
app = FastAPI()
|
14 |
|
15 |
-
# Define model and processor paths
|
16 |
-
RAG_MODEL = "vidore/colpali"
|
17 |
-
QWN_MODEL = "Qwen/Qwen2-VL-7B-Instruct"
|
18 |
-
QWN_PROCESSOR = "Qwen/Qwen2-VL-2B-Instruct"
|
19 |
-
|
20 |
# Load models and processors
|
21 |
-
RAG = RAGMultiModalModel.from_pretrained(RAG_MODEL)
|
22 |
|
23 |
-
|
24 |
QWN_MODEL,
|
25 |
torch_dtype=torch.bfloat16,
|
26 |
attn_implementation="flash_attention_2",
|
27 |
device_map="auto",
|
28 |
-
trust_remote_code=True
|
|
|
29 |
).cuda().eval()
|
30 |
|
31 |
-
|
32 |
|
33 |
# Define request model
|
34 |
class DocumentRequest(BaseModel):
|
35 |
text_query: str
|
36 |
|
37 |
-
# Define processing
|
38 |
-
def
|
|
|
|
|
|
|
|
|
|
|
39 |
messages = [
|
40 |
{
|
41 |
"role": "user",
|
42 |
"content": [
|
|
|
|
|
|
|
|
|
43 |
{
|
44 |
"type": "image",
|
45 |
"image": image,
|
46 |
},
|
47 |
-
{"type": "text", "text": text_query},
|
48 |
],
|
49 |
}
|
50 |
]
|
51 |
-
text =
|
52 |
messages, tokenize=False, add_generation_prompt=True
|
53 |
)
|
54 |
image_inputs, video_inputs = process_vision_info(messages)
|
55 |
-
inputs =
|
56 |
text=[text],
|
57 |
images=image_inputs,
|
58 |
videos=video_inputs,
|
@@ -60,26 +77,36 @@ def document_rag(text_query, image):
|
|
60 |
return_tensors="pt",
|
61 |
)
|
62 |
inputs = inputs.to("cuda")
|
63 |
-
generated_ids =
|
64 |
generated_ids_trimmed = [
|
65 |
out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
|
66 |
]
|
67 |
-
output_text =
|
68 |
generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
|
69 |
)
|
70 |
return output_text[0]
|
71 |
|
72 |
-
# Define API
|
73 |
@app.post("/process_document")
|
74 |
-
async def process_document(request: DocumentRequest, file: UploadFile = File(...)):
|
|
|
|
|
|
|
|
|
75 |
# Read and process the uploaded file
|
76 |
contents = await file.read()
|
77 |
image = Image.open(io.BytesIO(contents))
|
78 |
|
79 |
-
#
|
80 |
-
|
|
|
|
|
|
|
81 |
|
82 |
-
return {
|
|
|
|
|
|
|
83 |
|
84 |
if __name__ == "__main__":
|
85 |
import uvicorn
|
|
|
1 |
import os
|
2 |
+
from dotenv import load_dotenv
|
3 |
+
from fastapi import FastAPI, File, UploadFile, HTTPException, Header
|
4 |
from pydantic import BaseModel
|
5 |
+
from typing import Optional
|
6 |
import torch
|
7 |
from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
|
8 |
from qwen_vl_utils import process_vision_info
|
|
|
10 |
from PIL import Image
|
11 |
import io
|
12 |
|
13 |
+
# Load environment variables
|
14 |
+
load_dotenv()
|
15 |
+
|
16 |
+
# Access environment variables
|
17 |
+
HF_TOKEN = os.getenv("HF_TOKEN")
|
18 |
+
RAG_MODEL = os.getenv("RAG_MODEL", "vidore/colpali")
|
19 |
+
QWN_MODEL = os.getenv("QWN_MODEL", "Qwen/Qwen2-VL-7B-Instruct")
|
20 |
+
QWN_PROCESSOR = os.getenv("QWN_PROCESSOR", "Qwen/Qwen2-VL-2B-Instruct")
|
21 |
+
|
22 |
+
if not HF_TOKEN:
|
23 |
+
raise ValueError("HF_TOKEN not found in .env file")
|
24 |
+
|
25 |
# Initialize FastAPI app
|
26 |
app = FastAPI()
|
27 |
|
|
|
|
|
|
|
|
|
|
|
28 |
# Load models and processors
|
29 |
+
RAG = RAGMultiModalModel.from_pretrained(RAG_MODEL, use_auth_token=HF_TOKEN)
|
30 |
|
31 |
+
qwen_model = Qwen2VLForConditionalGeneration.from_pretrained(
|
32 |
QWN_MODEL,
|
33 |
torch_dtype=torch.bfloat16,
|
34 |
attn_implementation="flash_attention_2",
|
35 |
device_map="auto",
|
36 |
+
trust_remote_code=True,
|
37 |
+
use_auth_token=HF_TOKEN
|
38 |
).cuda().eval()
|
39 |
|
40 |
+
qwen_processor = AutoProcessor.from_pretrained(QWN_PROCESSOR, trust_remote_code=True, use_auth_token=HF_TOKEN)
|
41 |
|
42 |
# Define request model
|
43 |
class DocumentRequest(BaseModel):
|
44 |
text_query: str
|
45 |
|
46 |
+
# Define processing functions
|
47 |
+
def extract_text_with_colpali(image):
|
48 |
+
# Use ColPali (RAG) to extract text from the image
|
49 |
+
extracted_text = RAG.extract_text(image) # Assuming this method exists
|
50 |
+
return extracted_text
|
51 |
+
|
52 |
+
def process_with_qwen(query, extracted_text, image):
|
53 |
messages = [
|
54 |
{
|
55 |
"role": "user",
|
56 |
"content": [
|
57 |
+
{
|
58 |
+
"type": "text",
|
59 |
+
"text": f"Context: {extracted_text}\n\nQuery: {query}"
|
60 |
+
},
|
61 |
{
|
62 |
"type": "image",
|
63 |
"image": image,
|
64 |
},
|
|
|
65 |
],
|
66 |
}
|
67 |
]
|
68 |
+
text = qwen_processor.apply_chat_template(
|
69 |
messages, tokenize=False, add_generation_prompt=True
|
70 |
)
|
71 |
image_inputs, video_inputs = process_vision_info(messages)
|
72 |
+
inputs = qwen_processor(
|
73 |
text=[text],
|
74 |
images=image_inputs,
|
75 |
videos=video_inputs,
|
|
|
77 |
return_tensors="pt",
|
78 |
)
|
79 |
inputs = inputs.to("cuda")
|
80 |
+
generated_ids = qwen_model.generate(**inputs, max_new_tokens=100)
|
81 |
generated_ids_trimmed = [
|
82 |
out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
|
83 |
]
|
84 |
+
output_text = qwen_processor.batch_decode(
|
85 |
generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
|
86 |
)
|
87 |
return output_text[0]
|
88 |
|
89 |
+
# Define API endpoint
|
90 |
@app.post("/process_document")
|
91 |
+
async def process_document(request: DocumentRequest, file: UploadFile = File(...), x_api_key: Optional[str] = Header(None)):
|
92 |
+
# Check API key
|
93 |
+
if x_api_key != HF_TOKEN:
|
94 |
+
raise HTTPException(status_code=403, detail="Invalid API key")
|
95 |
+
|
96 |
# Read and process the uploaded file
|
97 |
contents = await file.read()
|
98 |
image = Image.open(io.BytesIO(contents))
|
99 |
|
100 |
+
# Extract text using ColPali
|
101 |
+
extracted_text = extract_text_with_colpali(image)
|
102 |
+
|
103 |
+
# Process the query with Qwen2, using both extracted text and image
|
104 |
+
result = process_with_qwen(request.text_query, extracted_text, image)
|
105 |
|
106 |
+
return {
|
107 |
+
"extracted_text": extracted_text,
|
108 |
+
"qwen_response": result
|
109 |
+
}
|
110 |
|
111 |
if __name__ == "__main__":
|
112 |
import uvicorn
|