Update app.py

app.py

@@ -8,12 +8,13 @@ import time
 app = FastAPI()
 
 # Create cache directory
-os.makedirs("./model_cache", exist_ok=True)
+cache_dir = "./model_cache"
+os.makedirs(cache_dir, exist_ok=True)
 
 # Track app status
 app_status = {
     "status": "initializing",
-    "model_name": "SmolLM2-135M-Instruct",
+    "model_name": "distilgpt2",
     "model_loaded": False,
     "tokenizer_loaded": False,
     "startup_time": time.time(),
@@ -21,22 +22,22 @@ app_status = {
 }
 
 # Load model and tokenizer once at startup
-model_name = "distilgpt2"  # change this to your own model
+model_name = "distilgpt2"
 try:
     # Try to load tokenizer
     app_status["status"] = "loading_tokenizer"
-    tokenizer = AutoTokenizer.from_pretrained(model_name, cache_dir="./model_cache", local_files_only=False)
+    tokenizer = AutoTokenizer.from_pretrained(model_name, cache_dir=cache_dir)
     app_status["tokenizer_loaded"] = True
     
-    # Try to load model
+    # Try to load model - with from_tf=True
     app_status["status"] = "loading_model"
-    model = AutoModelForCausalLM.from_pretrained(model_name, cache_dir="./model_cache", local_files_only=False)
+    model = AutoModelForCausalLM.from_pretrained(model_name, from_tf=True, cache_dir=cache_dir)
     app_status["model_loaded"] = True
     
     app_status["status"] = "ready"
 except Exception as e:
     error_msg = f"Error loading model or tokenizer: {str(e)}"
-    app_status["status"] = "error"
+    app_status["status"] = "limited_functionality"
     app_status["errors"].append(error_msg)
     print(error_msg)
 
@@ -48,7 +49,7 @@ class PromptRequest(BaseModel):
 async def generate_text(req: PromptRequest, response: Response):
     if app_status["status"] != "ready":
         response.status_code = status.HTTP_503_SERVICE_UNAVAILABLE
-        return {"error": "Model not ready", "status": app_status["status"]}
+        return {"error": "Model not ready", "status": app_status["status"], "details": app_status["errors"]}
     
     try:
         inputs = tokenizer(req.prompt, return_tensors="pt")
@@ -67,7 +68,7 @@ async def generate_text(req: PromptRequest, response: Response):
 
 @app.get("/")
 async def root():
-    return {"message": "API is running", "status": app_status["status"]}
+    return {"message": "API is responding", "status": app_status["status"]}
 
 @app.get("/status")
 async def get_status():