Upload app.py

app.py

@@ -5,6 +5,7 @@ import os
 import keras
 import numpy as np
 from transformers import RobertaTokenizer
+from huggingface_hub import hf_hub_download
 import uvicorn
 
 # Set Keras backend
@@ -23,6 +24,7 @@ MAX_LEN = 61
 
 # Hugging Face model repository
 HF_MODEL_ID = "Meshyboi/Multi-Emotion-Classification"
+MODEL_FILENAME = "roberta_emotion_model.keras"
 
 # Global variables for model and tokenizer
 model = None
@@ -42,15 +44,26 @@ class PredictionResponse(BaseModel):
     detected_emotions: List[str]
 
 def load_model():
+    """Load the trained model from Hugging Face"""
     global model
     try:
         if model is None:
-            model = keras.saving.load_model(f"hf://{HF_MODEL_ID}")
+            # Download the model file from Hugging Face
+            print(f"Downloading model file: {MODEL_FILENAME}")
+            model_path = hf_hub_download(
+                repo_id=HF_MODEL_ID,
+                filename=MODEL_FILENAME,
+                cache_dir=None  # Use default cache
+            )
+            print(f"Model downloaded to: {model_path}")
+            # Load the model from the downloaded file
+            model = keras.saving.load_model(model_path)
         return model
     except Exception as e:
         raise RuntimeError(f"Error loading model: {str(e)}")
 
 def load_tokenizer():
+    """Load the tokenizer from Hugging Face"""
     global tokenizer
     try:
         if tokenizer is None:
@@ -60,6 +73,7 @@ def load_tokenizer():
         raise RuntimeError(f"Error loading tokenizer: {str(e)}")
 
 def preprocess_text(text: str, tokenizer, max_len: int):
+    """Preprocess text for model input"""
     encoded = tokenizer.encode_plus(
         text,
         add_special_tokens=True,
@@ -72,6 +86,7 @@ def preprocess_text(text: str, tokenizer, max_len: int):
     return encoded['input_ids'], encoded['attention_mask']
 
 def predict_emotions(text: str, model, tokenizer):
+    """Predict emotions for given text"""
     input_ids, attention_mask = preprocess_text(text, tokenizer, MAX_LEN)
     predictions = model.predict([input_ids, attention_mask], verbose=0)
     return predictions[0]
@@ -86,6 +101,7 @@ async def startup_event():
 
 @app.get("/")
 async def root():
+    """Root endpoint"""
     return {
         "message": "Emotion Classification API",
         "version": "1.0.0",
@@ -98,6 +114,7 @@ async def root():
 
 @app.get("/health")
 async def health_check():
+    """Health check endpoint"""
     return {
         "status": "healthy",
         "model_loaded": model is not None,
@@ -106,6 +123,14 @@ async def health_check():
 
 @app.post("/predict", response_model=PredictionResponse)
 async def predict(request: PredictionRequest):
+    """
+    Predict emotions for the given text
+    
+    - **text**: Input text to analyze for emotions
+    
+    Returns:
+    - Dictionary with emotion scores and detected emotions
+    """
     if not request.text.strip():
         raise HTTPException(status_code=400, detail="Text cannot be empty")