Update app.py

app.py

@@ -64,10 +64,11 @@ def generate_responses(model, tokenizer, prompts):
     messages = [[{"role": "user", "content": message}] for message in prompts]
 
     texts = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-    print(texts)
+    print(texts[0])
+    
     model_inputs = tokenizer(texts, padding=True, truncation=True, max_length=512, return_tensors="pt").to(model.device)
 
-    print(tokenizer.batch_decode(model_inputs["input_ids"]))
+    print(tokenizer.batch_decode(model_inputs["input_ids"][0]))
 
     with torch.inference_mode():
         generated_ids = model.generate(
@@ -79,11 +80,11 @@ def generate_responses(model, tokenizer, prompts):
         )
     prompt_lengths = model_inputs["attention_mask"].sum(dim=1) - 1
     generated_ids = [output_ids[length:] for length, output_ids in zip(prompt_lengths, generated_ids)]
+    print(tokenizer.batch_decode(generated_ids[0], skip_special_tokens=False))
     responses = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
 
     return responses
 
-
 def classify_pairs(model, tokenizer, prompts, responses):
     texts = [prompt + "[SEP]" + response for prompt, response in zip(prompts, responses)]