Update app.py

app.py

@@ -25,7 +25,7 @@ def format_prompt(message, history):
 
 
 
-def chat_inf(system_prompt,prompt,history,client_choice):
+def chat_inf(system_prompt,prompt,history,client_choice,seed,temp,tokens,top_p,rep_p):
     #token max=8192
     client=clients[int(client_choice)-1]
     if not history:
@@ -35,12 +35,12 @@ def chat_inf(system_prompt,prompt,history,client_choice):
         hist_len=len(history)
         print(hist_len)
         
-    seed = random.randint(1,1111111111111111)
+    #seed = random.randint(1,1111111111111111)
     generate_kwargs = dict(
-        temperature=0.9,
-        max_new_tokens=6000,
-        top_p=0.95,
-        repetition_penalty=1.0,
+        temperature=temp,
+        max_new_tokens=tokens,
+        top_p=top_p,
+        repetition_penalty=rep_p,
         do_sample=True,
         seed=seed,
     )
@@ -57,6 +57,15 @@ def chat_inf(system_prompt,prompt,history,client_choice):
 
 def clear_fn():
     return None,None,None
+rand_val=random.randint(1,1111111111111111)
+def check_rand(inp,val):
+    if inp==True:
+        return gr.Slider(label="Seed", minimum=1, maximum=1111111111111111, value=random.randint(1,1111111111111111))
+    else:
+        return gr.Slider(label="Seed", minimum=1, maximum=1111111111111111, value=int(val))
+
+
+    
 with gr.Blocks() as app:
     gr.HTML("""<center><h1 style='font-size:xx-large;'>Google Gemma Models</h1><br><h3>running on Huggingface Inference Client</h3><br><h7>EXPERIMENTAL""")
     with gr.Group():
@@ -65,15 +74,26 @@ with gr.Blocks() as app:
             with gr.Column(scale=3):
                 inp = gr.Textbox(label="Prompt")
                 sys_inp = gr.Textbox(label="System Prompt (optional)")
-                btn = gr.Button("Chat")
-                
+                client_choice=gr.Dropdown(label="Models",type='index',choices=[c for c in models],value=models[0],interactive=True)
+                with gr.Row():
+                    with gr.Column(scale=2):
+                        btn = gr.Button("Chat")
+                    with gr.Column(scale=1):
+                        with gr.Group():
+                            stop_btn=gr.Button("Stop")
+                            clear_btn=gr.Button("Clear")                
             with gr.Column(scale=1):
                 with gr.Group():
-                    stop_btn=gr.Button("Stop")
-                    clear_btn=gr.Button("Clear")
-        client_choice=gr.Dropdown(label="Models",type='index',choices=[c for c in models],value=models[0],interactive=True)
+                    rand = gr.Checkbox(label="Random", value=True)
+                    seed=gr.Slider(label="Seed", minimum=1, maximum=1111111111111111,step=1, value=rand_val)
+                    tokens = gr.Slider(label="Max new tokens",value=6400,minimum=0,maximum=8000,step=64,interactive=True, visible=True,info="The maximum number of tokens")
+                    temp=gr.Slider(label="Temperature",step=0.01, minimum=0.01, maximum=1.0, value=0.9)
+                    top_p=gr.Slider(label="Top-P",step=0.01, minimum=0.01, maximum=1.0, value=0.9)
+                    rep_p=gr.Slider(label="Repetition Penalty",step=0.1, minimum=0.1, maximum=2.0, value=1.0)
+                    
+
     
-    go=btn.click(chat_inf,[sys_inp,inp,chat_b,client_choice],chat_b)
+    go=btn.click(check_rand,[rand,seed],seed).then(chat_inf,[sys_inp,inp,chat_b,client_choice,seed,temp,tokens,top_p,rep_p],chat_b)
     stop_btn.click(None,None,None,cancels=go)
     clear_btn.click(clear_fn,None,[inp,sys_inp,chat_b])
 app.queue(default_concurrency_limit=10).launch()