Update app.py

app.py

@@ -51,15 +51,15 @@ def run_gpt(
     stop_tokens,
     max_tokens,
     purpose,
-    **prompt_kwargs,
+    **prompt_kwargs,temperature=0.9, top_p=0.95, repetition_penalty=1.0,
 ):
     seed = random.randint(1,1111111111111111)
     print (seed)
     generate_kwargs = dict(
-        temperature=1.0,
+        temperature=temperature,
         max_new_tokens=1048,
-        top_p=0.99,
-        repetition_penalty=1.0,
+        top_p=top_p,
+        repetition_penalty=repetition_penalty,
         do_sample=True,
         seed=seed,
     )
@@ -87,19 +87,19 @@ def run_gpt(
     return resp
 
 
-def compress_history(purpose, task, history, directory):
+def compress_history(purpose, task, history, directory,temperature=0.9, top_p=0.95, repetition_penalty=1.0,):
     resp = run_gpt(
         COMPRESS_HISTORY_PROMPT,
         stop_tokens=["observation:", "task:", "action:", "thought:"],
         max_tokens=512,
         purpose=purpose,
         task=task,
-        history=history,
+        history=history,temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty,
     )
     history = "observation: {}\n".format(resp)
     return history
     
-def call_search(purpose, task, history, directory, action_input):
+def call_search(purpose, task, history, directory, action_input,temperature=0.9, top_p=0.95, repetition_penalty=1.0,):
     print("CALLING SEARCH")
     try:
         
@@ -119,14 +119,14 @@ def call_search(purpose, task, history, directory, action_input):
         history += "observation: {}'\n".format(e)
     return "MAIN", None, history, task
 
-def call_main(purpose, task, history, directory, action_input):
+def call_main(purpose, task, history, directory, action_input,temperature=0.9, top_p=0.95, repetition_penalty=1.0,):
     resp = run_gpt(
         ACTION_PROMPT,
         stop_tokens=["observation:", "task:"],
         max_tokens=1048,
         purpose=purpose,
         task=task,
-        history=history,
+        history=history,temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty,
     )
     lines = resp.strip().strip("\n").split("\n")
     for line in lines:
@@ -155,14 +155,14 @@ def call_main(purpose, task, history, directory, action_input):
     return "MAIN", None, history, task
 
 
-def call_set_task(purpose, task, history, directory, action_input):
+def call_set_task(purpose, task, history, directory, action_input,temperature=0.9, top_p=0.95, repetition_penalty=1.0,):
     task = run_gpt(
         TASK_PROMPT,
         stop_tokens=[],
         max_tokens=64,
         purpose=purpose,
         task=task,
-        history=history,
+        history=history,temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty,
     ).strip("\n")
     history += "observation: task has been updated to: {}\n".format(task)
     return "MAIN", None, history, task
@@ -179,7 +179,7 @@ NAME_TO_FUNC = {
 }
 
 
-def run_action(purpose, task, history, directory, action_name, action_input):
+def run_action(purpose, task, history, directory, action_name, action_input,temperature=0.9, top_p=0.95, repetition_penalty=1.0,):
     try:
         if "RESPONSE" in action_name or "COMPLETE" in action_name:
             action_name="COMPLETE"
@@ -198,13 +198,13 @@ def run_action(purpose, task, history, directory, action_name, action_input):
         assert action_name in NAME_TO_FUNC
     
         print("RUN: ", action_name, action_input)
-        return NAME_TO_FUNC[action_name](purpose, task, history, directory, action_input)
+        return NAME_TO_FUNC[action_name](purpose, task, history, directory, action_input,temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty,)
     except Exception as e:
         history += "observation: the previous command did not produce any useful output, I need to check the commands syntax, or use a different command\n"
 
         return "MAIN", None, history, task
 
-def run(purpose,history):
+def run(purpose,history,temperature=0.9, top_p=0.95, repetition_penalty=1.0,):
     
     #print(purpose)
     #print(hist)
@@ -233,7 +233,7 @@ def run(purpose,history):
             history,
             directory,
             action_name,
-            action_input,
+            action_input,temperature=temperature, top_p=top_p, repetition_penalty=repetition_penalty,
         )
         yield (history)
         #yield ("",[(purpose,history)])
@@ -295,17 +295,6 @@ def generate(
 
 
 additional_inputs=[
-    gr.Dropdown(
-        label="Agents",
-        choices=[s for s in agents],
-        value=agents[0],
-        interactive=True,
-        ),
-    gr.Textbox(
-        label="System Prompt",
-        max_lines=1,
-        interactive=True,
-    ),
     gr.Slider(
         label="Temperature",
         value=0.9,
@@ -315,16 +304,6 @@ additional_inputs=[
         interactive=True,
         info="Higher values produce more diverse outputs",
     ),
-
-    gr.Slider(
-        label="Max new tokens",
-        value=1048*10,
-        minimum=0,
-        maximum=1048*10,
-        step=64,
-        interactive=True,
-        info="The maximum numbers of new tokens",
-    ),
     gr.Slider(
         label="Top-p (nucleus sampling)",
         value=0.90,
@@ -383,6 +362,7 @@ iface.launch()
 '''
 gr.ChatInterface(
     fn=run,
+    additional_inputs=additional_inputs,
     chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
     title="Mixtral 46.7B\nMicro-Agent\nInternet Search <br> development test",
     examples=examples,