Update app.py

app.py

@@ -49,53 +49,52 @@ def process_text_pipeline(text):
     return "\n".join(processed_sentences)
 
 # ---------------- Additional Sentiment Models (No Sarcasm) ----------------
+# Pre-load tokenizers + models for safety
 additional_models = {
-    "siebert/sentiment-roberta-large-english": pipeline("sentiment-analysis", model="siebert/sentiment-roberta-large-english"),
-    "assemblyai/bert-large-uncased-sst2": AutoModelForSequenceClassification.from_pretrained("assemblyai/bert-large-uncased-sst2"),
-    "j-hartmann/sentiment-roberta-large-english-3-classes": pipeline("text-classification", model="j-hartmann/sentiment-roberta-large-english-3-classes", return_all_scores=True),
-    "cardiffnlp/twitter-xlm-roberta-base-sentiment": pipeline("sentiment-analysis", model="cardiffnlp/twitter-xlm-roberta-base-sentiment", tokenizer="cardiffnlp/twitter-xlm-roberta-base-sentiment"),
-    "sohan-ai/sentiment-analysis-model-amazon-reviews": DistilBertForSequenceClassification.from_pretrained("sohan-ai/sentiment-analysis-model-amazon-reviews")
+    "siebert/sentiment-roberta-large-english": {
+        "tokenizer": AutoTokenizer.from_pretrained("siebert/sentiment-roberta-large-english"),
+        "model": AutoModelForSequenceClassification.from_pretrained("siebert/sentiment-roberta-large-english")
+    },
+    "assemblyai/bert-large-uncased-sst2": {
+        "tokenizer": AutoTokenizer.from_pretrained("assemblyai/bert-large-uncased-sst2"),
+        "model": AutoModelForSequenceClassification.from_pretrained("assemblyai/bert-large-uncased-sst2")
+    },
+    "j-hartmann/sentiment-roberta-large-english-3-classes": {
+        "tokenizer": AutoTokenizer.from_pretrained("j-hartmann/sentiment-roberta-large-english-3-classes"),
+        "model": AutoModelForSequenceClassification.from_pretrained("j-hartmann/sentiment-roberta-large-english-3-classes")
+    },
+    "cardiffnlp/twitter-xlm-roberta-base-sentiment": {
+        "tokenizer": AutoTokenizer.from_pretrained("cardiffnlp/twitter-xlm-roberta-base-sentiment"),
+        "model": AutoModelForSequenceClassification.from_pretrained("cardiffnlp/twitter-xlm-roberta-base-sentiment")
+    },
+    "sohan-ai/sentiment-analysis-model-amazon-reviews": {
+        "tokenizer": DistilBertTokenizer.from_pretrained("distilbert-base-uncased"),
+        "model": DistilBertForSequenceClassification.from_pretrained("sohan-ai/sentiment-analysis-model-amazon-reviews")
+    }
 }
 
 def run_sentiment_with_selected_model(text, model_name):
-    if model_name == "siebert/sentiment-roberta-large-english":
-        result = additional_models[model_name](text)[0]
-        emoji = "✅" if result["label"].lower() == "positive" else "❌"
-        return f"{emoji} '{text}' -> {result['label']}"
-
-    elif model_name == "assemblyai/bert-large-uncased-sst2":
-        tokenizer = AutoTokenizer.from_pretrained(model_name)
-        model = additional_models[model_name]
-        tokens = tokenizer([text], return_tensors="pt", padding=True, truncation=True)
-        outputs = F.softmax(model(**tokens).logits, dim=1)
-        prob_pos = outputs[0][1].item()
-        prob_neg = outputs[0][0].item()
-        emoji = "✅" if prob_pos > prob_neg else "❌"
-        return f"{emoji} '{text}' -> Positive: {prob_pos:.2%}, Negative: {prob_neg:.2%}"
-
-    elif model_name == "j-hartmann/sentiment-roberta-large-english-3-classes":
-        results = additional_models[model_name](text)[0]
-        label_scores = {res['label']: res['score'] for res in results}
-        label = max(label_scores, key=label_scores.get)
-        emoji = "✅" if "positive" in label.lower() else "❌" if "negative" in label.lower() else "⚠️"
-        score_str = ", ".join([f"{k}: {v:.2%}" for k, v in label_scores.items()])
-        return f"{emoji} '{text}' -> {score_str}"
-
-    elif model_name == "cardiffnlp/twitter-xlm-roberta-base-sentiment":
-        result = additional_models[model_name](text)[0]
-        emoji = "✅" if result["label"].lower() == "positive" else "❌" if result["label"].lower() == "negative" else "⚠️"
-        return f"{emoji} '{text}' -> {result['label']}"
-
-    elif model_name == "sohan-ai/sentiment-analysis-model-amazon-reviews":
-        tokenizer = DistilBertTokenizer.from_pretrained("distilbert-base-uncased")
-        model = additional_models[model_name]
-        inputs = tokenizer(text, return_tensors="pt")
+    model_info = additional_models[model_name]
+    tokenizer = model_info["tokenizer"]
+    model = model_info["model"]
+
+    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
+    with torch.no_grad():
         outputs = model(**inputs)
-        label = "Positive" if outputs.logits.argmax().item() == 1 else "Negative"
-        emoji = "✅" if label == "Positive" else "❌"
-        return f"{emoji} '{text}' -> {label}"
 
-    return f"⚠️ Could not process with selected model."
+    logits = outputs.logits
+    probs = torch.nn.functional.softmax(logits, dim=-1)
+    pred = torch.argmax(probs, dim=-1).item()
+
+    # Get label from model config if available
+    if model.config.id2label:
+        label = model.config.id2label[pred]
+    else:
+        label = "Positive" if pred == 1 else "Negative"
+
+    emoji = "✅" if "positive" in label.lower() else "❌" if "negative" in label.lower() else "⚠️"
+    return f"{emoji} '{text}' -> {label}"
+
 
 # ---------------- Gradio UI ----------------
 background_css = """