Update main.py

main.py

@@ -11,10 +11,12 @@ app = FastAPI()
 
 MODEL = "cardiffnlp/twitter-roberta-base-sentiment-latest"
 
+# Load model and tokenizer
 tokenizer = AutoTokenizer.from_pretrained(MODEL)
 config = AutoConfig.from_pretrained(MODEL)
 model = AutoModelForSequenceClassification.from_pretrained(MODEL)
 
+# Preprocessing step for Twitter-style input
 def preprocess(text):
     tokens = []
     for t in text.split():
@@ -28,14 +30,34 @@ def preprocess(text):
 @app.post("/analyze")
 async def analyze(request: Request):
     data = await request.json()
-    text = preprocess(data.get("text", ""))
-    encoded_input = tokenizer(text, return_tensors='pt')
+    raw_text = data.get("text", "")
+
+    # Logging for debugging
+    print(f"Raw input: {raw_text}")
+
+    if not raw_text.strip():
+        return {"error": "Empty input text."}
+
+    text = preprocess(raw_text)
+    print(f"Preprocessed: {text}")
+
+    encoded_input = tokenizer(text, return_tensors='pt', truncation=True, padding=True)
+    print(f"Encoded input: {encoded_input.input_ids}")
+
     output = model(**encoded_input)
     scores = output[0][0].detach().numpy()
-    scores = softmax(scores)
+    probs = softmax(scores)
+
+    # Logging output
+    print(f"Raw scores: {scores}")
+    print(f"Softmax probs: {probs}")
+
     result = [
-        {"label": config.id2label[i], "score": round(float(scores[i]), 4)}
-        for i in scores.argsort()[::-1]
+        {"label": config.id2label[i], "score": round(float(probs[i]), 4)}
+        for i in probs.argsort()[::-1]
     ]
+
+    print(f"Result: {result}")
     return {"result": result}
 
+