Add Qwen3.5-0.8B OCR API

Dockerfile

@@ -0,0 +1,6 @@
+FROM python:3.12-slim
+WORKDIR /app
+COPY . .
+RUN pip install --no-cache-dir fastapi uvicorn torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu transformers pillow
+EXPOSE 7860
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,36 @@
+from fastapi import FastAPI, UploadFile, File, Form
+from pydantic import BaseModel
+from transformers import AutoModelForCausalLM, AutoProcessor
+import torch
+from PIL import Image
+import io
+
+app = FastAPI(title="Qwen3.5-0.8B OCR API")
+
+model_id = "Qwen/Qwen3.5-0.8B"
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="cpu")
+processor = AutoProcessor.from_pretrained(model_id)
+
+
+@app.post("/ocr")
+async def ocr(image: UploadFile = File(...), prompt: str = Form("Extract all text from this image accurately")):
+    img = Image.open(io.BytesIO(await image.read()))
+
+    messages = [{"role": "user", "content": [
+        {"type": "image", "image": img},
+        {"type": "text", "text": prompt}
+    ]}]
+
+    text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = processor(text, images=img, return_tensors="pt").to("cpu")
+
+    outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.7)
+    response = processor.decode(outputs[0], skip_special_tokens=True)
+
+    return {"extracted_text": response}
+
+
+# Simple test endpoint
+@app.get("/")
+async def root():
+    return {"message": "Qwen3.5-0.8B OCR API is running! Use POST /ocr with image file."}

requirements.txt

@@ -0,0 +1,5 @@
+fastapi
+uvicorn
+torch --index-url https://download.pytorch.org/whl/cpu
+transformers
+pillow