Update app.py

app.py

@@ -1,160 +1,40 @@
 import gradio as gr
-from gradio_client import Client
-from huggingface_hub import InferenceClient
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
 import random
-ss_client = Client("https://omnibus-html-image-current-tab.hf.space/")
 
-models=[
-    "google/gemma-7b",
-    "google/gemma-7b-it",
-    "google/gemma-2b",
-    "google/gemma-2b-it"
-]
-clients=[
-InferenceClient(models[0]),
-InferenceClient(models[1]),
-InferenceClient(models[2]),
-InferenceClient(models[3]),
-]
-
-VERBOSE=False
-
-def load_models(inp):
-    if VERBOSE==True:    
-        print(type(inp))
-        print(inp)
-        print(models[inp])
-    #client_z.clear()
-    #client_z.append(InferenceClient(models[inp]))
-    return gr.update(label=models[inp])
-
-def format_prompt(message, history, cust_p):
-    prompt = ""
-    if history:
-        for user_prompt, bot_response in history:
-            prompt += f"<start_of_turn>user{user_prompt}<end_of_turn>"
-            prompt += f"<start_of_turn>model{bot_response}<end_of_turn>"
-            if VERBOSE==True:
-                print(prompt)
-    #prompt += f"<start_of_turn>user\n{message}<end_of_turn>\n<start_of_turn>model\n"
-    prompt+=cust_p.replace("USER_INPUT",message)
-    return prompt
-
-def chat_inf(system_prompt,prompt,history,memory,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem,cust_p):
-    #token max=8192
-    print(client_choice)
-    hist_len=0
-    client=clients[int(client_choice)-1]
-    if not history:
-        history = []
-        hist_len=0
-    if not memory:
-        memory = []
-        mem_len=0        
-    if memory:
-        for ea in memory[0-chat_mem:]:
-            hist_len+=len(str(ea))
-    in_len=len(system_prompt+prompt)+hist_len
-
-    if (in_len+tokens) > 8000:
-        history.append((prompt,"Wait, that's too many tokens, please reduce the 'Chat Memory' value, or reduce the 'Max new tokens' value"))
-        yield history,memory
-    else:
-        generate_kwargs = dict(
-            temperature=temp,
-            max_new_tokens=tokens,
-            top_p=top_p,
-            repetition_penalty=rep_p,
-            do_sample=True,
-            seed=seed,
-        )
-        if system_prompt:
-            formatted_prompt = format_prompt(f"{system_prompt}, {prompt}", memory[0-chat_mem:],cust_p)
-        else:
-            formatted_prompt = format_prompt(prompt, memory[0-chat_mem:],cust_p)
-        stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
-        output = ""
-        for response in stream:
-            output += response.token.text
-            yield [(prompt,output)],memory
-        history.append((prompt,output))
-        memory.append((prompt,output))
-        yield history,memory
-        
-    if VERBOSE==True:
-        print("\n######### HIST "+str(in_len))
-        print("\n######### TOKENS "+str(tokens))        
+models = ["microsoft/DialoGPT-medium", "facebook/opt-125m"]
+tokenizers = {name: AutoTokenizer.from_pretrained(name) for name in models}
+clients = {name: AutoModelForCausalLM.from_pretrained(name).to('cpu') for name in models}
+ss_client=AutoModelForCausalLM.from_pretrained("nchlt/omnibus-image-current-tab").to('cuda')
+
+def load_models(choice):
+    return clients[choice],tokenizers[choice]
+
+def chat_inf(sys, inp, chat, mem, cli, seed, temp, tokens, top_p, rep_p, chat_mem, custom_prompt):
+    torch.manual_seed(int(seed))
+    if not sys:
+        sys = "<|startoftext|>"
+    if inp is None:
+        return [],[]
+    history=[(inp,chat)]
+    chat+=[inp]
+    response = cli.generate(torch.tensor([tokenizers[cli].encode(f'{sys}: {inp}\n') for inp in chat]), max_length=int(tokens), temperature=float(temp), top_p=float(top_p), do_sample=True, repetition_penalty=float(rep_p))
+    res = tokenizers[cli].decode(response[:, -1])
+    chat+=[res]
+    
+    custom_prompt.text = "\n".join([f"{i}: {inp} <--> {res}" for i,(inp,res) in enumerate(history[::-1][:chat_mem])])
+    return res, chat
 
-def get_screenshot(chat: list,height=5000,width=600,chatblock=[],theme="light",wait=3000,header=True):
-    print(chatblock)
-    tog = 0
-    if chatblock:
-        tog = 3
-    result = ss_client.predict(str(chat),height,width,chatblock,header,theme,wait,api_name="/run_script")
-    out = f'https://omnibus-html-image-current-tab.hf.space/file={result[tog]}'
-    print(out)
-    return out
+def get_screenshot(cli, im_height, im_width, chatblock, theme, wait_time):
+    chat = cli.generate(torch.tensor([tokenizers[cli].encode('<|startoftext|>: '+'\n'.join([inp for i,(inp,res) in enumerate(history[::-1][:chatblock]) if not i%2])) for _ in range(5)]), max_length=int(tokens), temperature=float(temp), top_p=float(top_p), do_sample=True, repetition_penalty=float(rep_p))
+    return tokenizers[cli].decode(response[:, -1])
 
 def clear_fn():
-    return None,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:
-    memory=gr.State()
-    gr.HTML("""<center><h1 style='font-size:xx-large;'>Google Gemma Models</h1><br><h3>running on Huggingface Inference Client</h3><br><h7>EXPERIMENTAL""")
-    chat_b = gr.Chatbot(height=500)
-    with gr.Group():
-        with gr.Row():
-            with gr.Column(scale=3):
-                inp = gr.Textbox(label="Prompt")
-                sys_inp = gr.Textbox(label="System Prompt (optional)")
-                with gr.Accordion("Prompt Format",open=False):
-                    custom_prompt=gr.Textbox(label="Modify Prompt Format", info="For testing purposes. 'USER_INPUT' is where 'SYSTEM_PROMPT, PROMPT' will be placed", lines=3,value="<start_of_turn>userUSER_INPUT<end_of_turn><start_of_turn>model")                
-                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")                
-                client_choice=gr.Dropdown(label="Models",type='index',choices=[c for c in models],value=models[0],interactive=True)
-            with gr.Column(scale=1):
-                with gr.Group():
-                    rand = gr.Checkbox(label="Random Seed", value=True)
-                    seed=gr.Slider(label="Seed", minimum=1, maximum=1111111111111111,step=1, value=rand_val)
-                    tokens = gr.Slider(label="Max new tokens",value=1600,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=0.49)
-                    top_p=gr.Slider(label="Top-P",step=0.01, minimum=0.01, maximum=1.0, value=0.49)
-                    rep_p=gr.Slider(label="Repetition Penalty",step=0.01, minimum=0.1, maximum=2.0, value=0.99)
-                    chat_mem=gr.Number(label="Chat Memory", info="Number of previous chats to retain",value=4)
-        with gr.Accordion(label="Screenshot",open=False):
-            with gr.Row():
-                with gr.Column(scale=3):
-                    im_btn=gr.Button("Screenshot")
-                    img=gr.Image(type='filepath')
-                with gr.Column(scale=1):
-                    with gr.Row():
-                        im_height=gr.Number(label="Height",value=5000)
-                        im_width=gr.Number(label="Width",value=500)
-                    wait_time=gr.Number(label="Wait Time",value=3000)
-                    theme=gr.Radio(label="Theme", choices=["light","dark"],value="light")
-                    chatblock=gr.Dropdown(label="Chatblocks",info="Choose specific blocks of chat",choices=[c for c in range(1,40)],multiselect=True)
-
-    
-    client_choice.change(load_models,client_choice,[chat_b])
-    app.load(load_models,client_choice,[chat_b])
-    
-    im_go=im_btn.click(get_screenshot,[chat_b,im_height,im_width,chatblock,theme,wait_time],img)
-    
-    chat_sub=inp.submit(check_rand,[rand,seed],seed).then(chat_inf,[sys_inp,inp,chat_b,memory,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem,custom_prompt],[chat_b,memory])
-    go=btn.click(check_rand,[rand,seed],seed).then(chat_inf,[sys_inp,inp,chat_b,memory,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem,custom_prompt],[chat_b,memory])
+    inp.value = ""
+    sys_inp.value = ""
+    chat_b.value = []
+    memory.value = None
     
-    stop_btn.click(None,None,None,cancels=[go,im_go,chat_sub])
-    clear_btn.click(clear_fn,None,[inp,sys_inp,chat_b,memory])
+im_go=im_btn.click(get_screenshot,[chat_b,im_height,im_width,chatblock,theme,wait_time],img)
 app.queue(default_concurrency_limit=10).launch()