Upload folder using huggingface_hub

Dockerfile

@@ -0,0 +1,25 @@
+
+# Lightweight Python base
+FROM python:3.12-slim
+
+WORKDIR /app
+COPY . /app
+
+# Create writable cache folder
+RUN mkdir -p /app/model_cache && chmod -R 777 /app/model_cache
+
+# Environment variables for Hugging Face cache
+ENV HF_HOME=/app/model_cache
+ENV TRANSFORMERS_CACHE=/app/model_cache
+ENV HF_DATASETS_CACHE=/app/model_cache
+ENV HF_METRICS_CACHE=/app/model_cache
+
+# Install dependencies
+RUN apt-get update && apt-get install -y git
+RUN pip install --upgrade pip
+RUN pip install --no-cache-dir -r requirements.txt
+
+EXPOSE 7860
+
+# Run FastAPI server
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,53 @@
+
+import os
+from fastapi import FastAPI
+from pydantic import BaseModel
+from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
+import torch
+
+# ✅ Create writable model cache directory
+os.makedirs("/app/model_cache", exist_ok=True)
+os.environ["HF_HOME"] = "/app/model_cache"
+os.environ["TRANSFORMERS_CACHE"] = "/app/model_cache"
+os.environ["HF_DATASETS_CACHE"] = "/app/model_cache"
+os.environ["HF_METRICS_CACHE"] = "/app/model_cache"
+
+MODEL_REPO = "kar137/sambodhan-urgency-classifier"
+device = 0 if torch.cuda.is_available() else -1
+
+# ✅ Load model and tokenizer safely
+tokenizer = AutoTokenizer.from_pretrained(MODEL_REPO, cache_dir="/app/model_cache")
+model = AutoModelForSequenceClassification.from_pretrained(MODEL_REPO, cache_dir="/app/model_cache")
+
+classifier = pipeline(
+    "text-classification",
+    model=model,
+    tokenizer=tokenizer,
+    device=device,
+    return_all_scores=True
+)
+
+LABELS = ["NORMAL", "URGENT", "HIGHLY URGENT"]
+
+app = FastAPI(title="Sambodhan Urgency Classifier API", version="2.0.3")
+
+class TextInput(BaseModel):
+    text: str
+
+@app.post("/predict")
+async def predict(input_data: TextInput):
+    text = input_data.text.strip()
+    if not text:
+        return {"error": "Empty input"}
+
+    results = classifier(text)[0]
+    top = max(results, key=lambda x: x["score"])
+    return {
+        "label": top["label"],
+        "confidence": round(top["score"], 4),
+        "scores": {r["label"]: round(r["score"], 4) for r in results},
+    }
+
+@app.get("/")
+def root():
+    return {"message": "✅ Sambodhan Urgency Classifier API running successfully!"}

requirements.txt

@@ -0,0 +1,7 @@
+fastapi
+uvicorn
+transformers
+torch
+huggingface-hub
+protobuf
+sentencepiece