Delete app_test.py

app_test.py

@@ -1,52 +0,0 @@
-from transformers import pipeline
-
-# Load pipelines
-pipe_bert = pipeline("text-classification", model="distilbert/distilbert-base-uncased-finetuned-sst-2-english")
-pipe_roberta = pipeline("text-classification", model="cardiffnlp/twitter-roberta-base-sentiment")
-
-# Label mapping for RoBERTa
-roberta_label_mapping_dict = {
-    'LABEL_2': 'Positive',
-    'LABEL_1': 'Neutral',
-    'LABEL_0': 'Negative'
-}
-
-# Sample input - top 100 review
-top_n = 10 
-reviews = train_data['review'][:top_n]
-sentiments = train_data['sentiment'][:top_n]
-data_to_test = dict(zip(reviews, sentiments))
-
-# Print header
-print(f"{'Original':<10} | {'DistilBERT':<10} | {'RoBERTa':<10}")
-
-# Track accuracy
-bert_correct = 0
-roberta_correct = 0
-total = len(data_to_test)
-
-for text, true_label in data_to_test.items():
-    pred_bert = pipe_bert(text)[0]
-    pred_roberta = pipe_roberta(text, truncation=True)[0]
-
-    # Normalize labels
-    original = true_label.strip().capitalize()
-    bert = pred_bert["label"].capitalize()
-    roberta = roberta_label_mapping_dict.get(pred_roberta["label"], "Unknown")
-
-    # Accuracy check
-    if bert == original:
-        bert_correct += 1
-    if roberta == original:
-        roberta_correct += 1
-
-    # Print results
-    print(f"{original:<10} | {bert:<10} | {roberta:<10}")
-
-# Calculate and print accuracy
-bert_acc = (bert_correct / total) * 100
-roberta_acc = (roberta_correct / total) * 100
-
-print(f"\nAccuracy:")
-print(f"DistilBERT: {bert_acc:.2f}%")
-print(f"RoBERTa   : {roberta_acc:.2f}%")