Update app.py

app.py

@@ -537,10 +537,11 @@ def generate_text_with_probs(initial_context, context_text , top_p, max_length,
 
             ##print(generated_text)
             initial_context = generated_text
-            input_ids = tokenizer.encode(generated_text, return_tensors="pt").to(device='mps')
+            #input_ids = tokenizer.encode(generated_text, return_tensors="pt").to(device='mps')
+            input_ids = tokenizer.encode(generated_text, return_tensors="pt").to(device='cpu')
 
-            use_ngram += 1
 
+            use_ngram += 1
 
           else:
 
@@ -576,7 +577,7 @@ def generate_text_with_probs(initial_context, context_text , top_p, max_length,
 
     generated_text = tokenizer.decode(input_ids[0], skip_special_tokens=True)
 
-    total = use_llm + use_llm_back_up + use_ngram
+    #total = use_llm + use_llm_back_up + use_ngram
 
     ##print(f"total: {use_llm} ({(use_llm / total) * 100:.2f}%)")
     ##print(f"use_llms: {use_llm_back_up} ({(use_llm_back_up / total) * 100:.2f}%)")