simple

app.py

@@ -23,17 +23,15 @@ import gradio as gr
 def chat(message, history=[]):
     new_user_input_ids = tokenizer.encode(message+tokenizer.eos_token, return_tensors='pt')
     if  len(history) > 0:
-        last_set_of_ids = history[len(history)-1][3]
+        last_set_of_ids = history[len(history)-1][2]
         bot_input_ids = torch.cat([last_set_of_ids, new_user_input_ids], dim=-1) 
     else:
         print("HERE WE GO! ", new_user_input_ids)
         bot_input_ids = new_user_input_ids
-    chat_history_ids = model.generate(bot_input_ids, max_length=500, pad_token_id=tokenizer.eos_token_id)
+    chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id)
     response_ids = chat_history_ids[:, bot_input_ids.shape[-1]:][0]
     response = tokenizer.decode(response_ids, skip_special_tokens=True)
-    # response = tokenizer.decode(response_ids).replace("<|endoftext|>", "")
-    bot_input_ids = torch.cat([bot_input_ids, response_ids], dim=-1)
-    history.push((message, response, bot_input_ids))
+    history.push((message, response, chat_history_id))
     return response, history, feedback(message)