Fix: set HF cache to /app/cache to avoid permission error

Dockerfile

@@ -5,8 +5,10 @@ RUN pip install --no-cache-dir -r requirements.txt
 
 COPY app.py .
 
-# Hugging Face model cache
+
+# Create cache dir and set env vars so HF/Transformers don't use `/`
 RUN mkdir -p /app/cache
+ENV HF_HOME=/app/cache
 ENV TRANSFORMERS_CACHE=/app/cache
 
 EXPOSE 7860

app.py

@@ -3,6 +3,9 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 from peft import PeftModel
 import torch
 from pydantic import BaseModel
+import os
+os.environ["HF_HOME"] = "/app/cache"
+os.environ["TRANSFORMERS_CACHE"] = "/app/cache"
 
 app = FastAPI()
 
@@ -23,6 +26,8 @@ def generate_text(request: GenerationRequest):
     inputs = tokenizer(request.prompt, return_tensors="pt")
     outputs = model.generate(**inputs, max_length=request.max_length)
     text = tokenizer.decode(outputs[0], skip_special_tokens=True, do_sample=True, top_k=20)
+    print("Input Prompt:", request.prompt)
+    print("Generated Text:", text)
     return {"generated_text": text}