revert

app.py

@@ -231,7 +231,7 @@ def predict_text_streaming(text):
         for model_name in ['E5 Classifier', 'E5-Instruct Classifier']:
             start_time = time.time()
             model = models[model_name]
-            embedding_2d = np.asarray(e5_embedding).reshape(1, -1)
+            embedding_2d = e5_embedding.reshape(1, -1)
             prediction = model.predict(embedding_2d)[0]
             probabilities = model.predict_proba(embedding_2d)[0]
             confidence = max(probabilities)
@@ -251,7 +251,7 @@ def predict_text_streaming(text):
         for model_name in ['Azure Classifier', 'Azure KNN Classifier']:
             start_time = time.time()
             model = models[model_name]
-            embedding_2d = np.asarray(azure_embedding).reshape(1, -1)
+            embedding_2d = azure_embedding.reshape(1, -1)
             prediction = model.predict(embedding_2d)[0]
             probabilities = model.predict_proba(embedding_2d)[0]
             confidence = max(probabilities)
@@ -269,7 +269,7 @@ def predict_text_streaming(text):
         yield format_progress(80, "Processing ModernBERT RF Classifier..."), format_results(results)
         modernbert_embedding, embed_time = generate_modernbert_embedding(text)
         model = models['ModernBERT RF Classifier']
-        embedding_2d = np.asarray(modernbert_embedding).reshape(1, -1)
+        embedding_2d = modernbert_embedding.reshape(1, -1)
         prediction = model.predict(embedding_2d)[0]
         probabilities = model.predict_proba(embedding_2d)[0]
         confidence = max(probabilities)
@@ -287,7 +287,7 @@ def predict_text_streaming(text):
         yield format_progress(95, "Processing GTE Classifier..."), format_results(results)
         gte_embedding, embed_time = generate_gte_embedding(text)
         model = models['GTE Classifier']
-        embedding_2d = np.asarray(gte_embedding).reshape(1, -1)
+        embedding_2d = gte_embedding.reshape(1, -1)
         prediction = model.predict(embedding_2d)[0]
         probabilities = model.predict_proba(embedding_2d)[0]
         confidence = max(probabilities)

requirements.txt

@@ -2,6 +2,6 @@ gradio>=4.0.0
 fasttext>=0.9.2
 numpy==1.26.4
 torch>=2.0.0
-transformers>=4.30.0
+transformers==4.48.0.dev0
 openai>=1.0.0
 scikit-learn