Tonic commited on
Commit
06271b6
1 Parent(s): 6ab3fad

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +9 -9
app.py CHANGED
@@ -54,23 +54,23 @@ class FalconChatBot:
54
  def predict(self, system_prompt, user_message, assistant_message, history, temperature, max_new_tokens, top_p, repetition_penalty):
55
 
56
  # Process the history to remove special commands
57
- processed_history = self.process_history(history)
58
 
59
  # Combine the user and assistant messages into a conversation
60
- conversation = f"{system_prompt}\nFalcon: {assistant_message if assistant_message else ''} User: {user_message}\nFalcon:\n"
61
 
62
  # Encode the conversation using the tokenizer
63
- input_ids = tokenizer.encode(conversation, return_tensors="pt", add_special_tokens=False)
64
 
65
  # Generate a response using the Falcon model
66
- response_text = peft_model.generate(input_ids=input_ids, max_length=max_length, use_cache=True, early_stopping=True, bos_token_id=peft_model.config.bos_token_id, eos_token_id=peft_model.config.eos_token_id, pad_token_id=peft_model.config.eos_token_id, temperature=0.4, do_sample=True)
67
 
68
  # Generate the formatted conversation in Falcon message format
69
- conversation = f"{system_prompt}\n"
70
- for message in processed_history:
71
- user_message = message["user"]
72
- assistant_message = message["assistant"]
73
- conversation += f"Falcon:{' ' + assistant_message if assistant_message else ''} User: {user_message}\n Falcon:\n"
74
 
75
  return response_text
76
 
 
54
  def predict(self, system_prompt, user_message, assistant_message, history, temperature, max_new_tokens, top_p, repetition_penalty):
55
 
56
  # Process the history to remove special commands
57
+ processed_history = self.process_history(history)
58
 
59
  # Combine the user and assistant messages into a conversation
60
+ conversation = f"{system_prompt}\nFalcon: {assistant_message if assistant_message else ''} User: {user_message}\nFalcon:\n"
61
 
62
  # Encode the conversation using the tokenizer
63
+ input_ids = tokenizer.encode(conversation, return_tensors="pt", add_special_tokens=False)
64
 
65
  # Generate a response using the Falcon model
66
+ response_text = peft_model.generate(input_ids=input_ids, max_length=max_length, use_cache=True, early_stopping=True, bos_token_id=peft_model.config.bos_token_id, eos_token_id=peft_model.config.eos_token_id, pad_token_id=peft_model.config.eos_token_id, temperature=0.4, do_sample=True)
67
 
68
  # Generate the formatted conversation in Falcon message format
69
+ conversation = f"{system_prompt}\n"
70
+ for message in processed_history:
71
+ user_message = message["user"]
72
+ assistant_message = message["assistant"]
73
+ conversation += f"Falcon:{' ' + assistant_message if assistant_message else ''} User: {user_message}\n Falcon:\n"
74
 
75
  return response_text
76