#uses HuggingFace pipeline to classify text sentiment (positive, neutral, negative) with FinBERT. from transformers import pipeline # Load FinBERT financial sentiment pipeline sentiment_pipeline = pipeline( "sentiment-analysis", model="ProsusAI/finbert", top_k=None ) def analyze_sentiment(text): try: result = sentiment_pipeline(text[:512])[0] # limit size for tokenizer return result["label"].lower(), round(result["score"], 3) except Exception as e: return "error", 0.0