Spaces:
Sleeping
Sleeping
Commit
·
5e9e76c
1
Parent(s):
bb18f6f
Add batching using vllm
Browse files- main.py +44 -0
- requirements.txt +2 -1
main.py
CHANGED
@@ -77,6 +77,9 @@ import logging
|
|
77 |
from typing import List
|
78 |
import time
|
79 |
import numpy as np
|
|
|
|
|
|
|
80 |
|
81 |
# Configure logging
|
82 |
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
|
@@ -97,6 +100,7 @@ app = FastAPI()
|
|
97 |
# Global variables for model and processor
|
98 |
model = None
|
99 |
processor = None
|
|
|
100 |
|
101 |
def load_model():
|
102 |
"""Load model and processor when needed"""
|
@@ -114,6 +118,15 @@ def load_model():
|
|
114 |
processor = PaliGemmaProcessor.from_pretrained(model_id)
|
115 |
logger.info("Model loaded successfully")
|
116 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
117 |
def clean_memory():
|
118 |
"""Force garbage collection and clear CUDA cache"""
|
119 |
gc.collect()
|
@@ -165,6 +178,37 @@ async def extract_text(background_tasks: BackgroundTasks, file: UploadFile = Fil
|
|
165 |
logger.error(f"Error processing image: {str(e)}")
|
166 |
return {"error": str(e)}
|
167 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
168 |
@app.post("/batch_extract_text")
|
169 |
async def batch_extract_text(batch_size:int, background_tasks: BackgroundTasks, files: List[UploadFile] = File(...)):
|
170 |
"""Extract text from multiple images with batching"""
|
|
|
77 |
from typing import List
|
78 |
import time
|
79 |
import numpy as np
|
80 |
+
from vllm import LLM, SamplingParams
|
81 |
+
import torch._dynamo
|
82 |
+
torch._dynamo.config.suppress_errors = True
|
83 |
|
84 |
# Configure logging
|
85 |
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
|
|
|
100 |
# Global variables for model and processor
|
101 |
model = None
|
102 |
processor = None
|
103 |
+
llm = None
|
104 |
|
105 |
def load_model():
|
106 |
"""Load model and processor when needed"""
|
|
|
118 |
processor = PaliGemmaProcessor.from_pretrained(model_id)
|
119 |
logger.info("Model loaded successfully")
|
120 |
|
121 |
+
def load_vllm_model():
|
122 |
+
global llm
|
123 |
+
if llm is None:
|
124 |
+
llm = LLM(
|
125 |
+
model="google/paligemma2-3b-mix-448",
|
126 |
+
trust_remote_code=True,
|
127 |
+
max_model_len=4096,
|
128 |
+
dtype="float16",
|
129 |
+
)
|
130 |
def clean_memory():
|
131 |
"""Force garbage collection and clear CUDA cache"""
|
132 |
gc.collect()
|
|
|
178 |
logger.error(f"Error processing image: {str(e)}")
|
179 |
return {"error": str(e)}
|
180 |
|
181 |
+
@app.post("/batch_extract_text_vllm")
|
182 |
+
async def batch_extract_text_vllm(background_tasks: BackgroundTasks, files: List[UploadFile] = File(...)):
|
183 |
+
try:
|
184 |
+
start_time = time.time()
|
185 |
+
load_vllm_model()
|
186 |
+
results = []
|
187 |
+
sampling_params = SamplingParams(temperature=0.0,max_tokens=32)
|
188 |
+
# Load images
|
189 |
+
images = []
|
190 |
+
for file in files:
|
191 |
+
image_data = await file.read()
|
192 |
+
img = Image.open(BytesIO(image_data)).convert("RGB")
|
193 |
+
images.append(img)
|
194 |
+
for image in images:
|
195 |
+
inputs = {
|
196 |
+
"prompt": "ocr",
|
197 |
+
"multi_modal_data": {
|
198 |
+
"image": image
|
199 |
+
},
|
200 |
+
}
|
201 |
+
outputs = llm.generate(inputs, sampling_params)
|
202 |
+
for o in outputs:
|
203 |
+
generated_text = o.outputs[0].text
|
204 |
+
results.append(generated_text)
|
205 |
+
|
206 |
+
logger.info(f"vLLM Batch processing completed in {time.time() - start_time:.2f} seconds")
|
207 |
+
return {"extracted_texts": results}
|
208 |
+
except Exception as e:
|
209 |
+
logger.error(f"Error in batch processing vLLM: {str(e)}")
|
210 |
+
return {"error": str(e)}
|
211 |
+
|
212 |
@app.post("/batch_extract_text")
|
213 |
async def batch_extract_text(batch_size:int, background_tasks: BackgroundTasks, files: List[UploadFile] = File(...)):
|
214 |
"""Extract text from multiple images with batching"""
|
requirements.txt
CHANGED
@@ -7,4 +7,5 @@ transformers
|
|
7 |
torch
|
8 |
accelerate
|
9 |
pillow
|
10 |
-
python-multipart
|
|
|
|
7 |
torch
|
8 |
accelerate
|
9 |
pillow
|
10 |
+
python-multipart
|
11 |
+
vllm
|