app.py

import gradio as gr
from gradio_client import Client
from huggingface_hub import InferenceClient
import random
from datetime import datetime
from models import models
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",
    "openchat/openchat-3.5-0106",
    "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
    "mistralai/Mixtral-8x7B-Instruct-v0.1",
    "JunRyeol/jr_model",
]
'''


def test_models():
    log_box=[]
    for model in models:
        start_time = datetime.now()
        try:

            generate_kwargs = dict(
                temperature=0.9,
                max_new_tokens=128,
                top_p=0.9,
                repetition_penalty=1.0,
                do_sample=True,
                seed=111111111,
            )
          
            print(f'trying: {model}\n')
            client= InferenceClient(model)
            outp=""
            stream=client.text_generation("What is a cat", **generate_kwargs, stream=True, details=True, return_full_text=True)
            for response in stream:
                outp += response.token.text
                print (outp)            
            time_delta = datetime.now() - start_time
            count=time_delta.total_seconds()
            #if time_delta.total_seconds() >= 180:
            log = {"Model":model,"Status":"Success","Output":outp, "Time":count}
            print(f'{log}\n')
            log_box.append(log)
        except Exception as e:
            time_delta = datetime.now() - start_time
            count=time_delta.total_seconds()

            log = {"Model":model,"Status":"Error","Output":e,"Time":count}
            print(f'{log}\n')
            log_box.append(log)
        yield log_box

def format_prompt_default(message, history,cust_p):
    prompt = ""
    if history:
        #<start_of_turn>userHow does the brain work?<end_of_turn><start_of_turn>model
        for user_prompt, bot_response in history:
            prompt += f"{user_prompt}\n"
            print(prompt)
            prompt += f"{bot_response}\n"
            print(prompt)
    #prompt += f"{message}\n"
    prompt+=cust_p.replace("USER_INPUT",message)
    return prompt

def format_prompt_gemma(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 format_prompt_openc(message, history,cust_p):
    #prompt = "GPT4 Correct User: "
    prompt=""
    if history:
        #<start_of_turn>userHow does the brain work?<end_of_turn><start_of_turn>model
        for user_prompt, bot_response in history:
            prompt += f"{user_prompt}"
            prompt += f"<|end_of_turn|>"
            prompt += f"GPT4 Correct Assistant: "
            prompt += f"{bot_response}"
            prompt += f"<|end_of_turn|>"
            print(prompt)
    #GPT4 Correct User: Hello<|end_of_turn|>GPT4 Correct Assistant: 
    prompt+=cust_p.replace("USER_INPUT",message)
    return prompt
   
def format_prompt_mixtral(message, history,cust_p):
    prompt = "<s>"
    if history:
        for user_prompt, bot_response in history:
            prompt += f"[INST] {user_prompt} [/INST]"
            prompt += f" {bot_response}</s> "
    #prompt += f"[INST] {message} [/INST]"
    prompt+=cust_p.replace("USER_INPUT",message)    
    return prompt

def format_prompt_choose(message, history, cust_p, model_name):
    if "gemma" in models[model_name].lower():
        return format_prompt_gemma(message,history,cust_p)
    if "mixtral" in models[model_name].lower():
        return format_prompt_mixtral(message,history,cust_p)
    if "openchat" in models[model_name].lower():
        return format_prompt_openc(message,history,cust_p)        
    else:
        return format_prompt_default(message,history,cust_p)

def load_models(inp):
    print(type(inp))
    print(inp)
    print(models[inp])
    model_state= InferenceClient(models[inp])
    out_box=gr.update(label=models[inp])
    if "gemma" in models[inp].lower():
        prompt_out="<start_of_turn>userUSER_INPUT<end_of_turn><start_of_turn>model"
        return out_box,prompt_out, model_state
    if "mixtral" in models[inp].lower():
        prompt_out="[INST] USER_INPUT [/INST]"
        return out_box,prompt_out, model_state
    if "openchat" in models[inp].lower():
        prompt_out="GPT4 Correct User: USER_INPUT<|end_of_turn|>GPT4 Correct Assistant: "
        return out_box,prompt_out, model_state   
    else:
        prompt_out="USER_INPUT\n"
        return out_box,prompt_out, model_state
    

VERBOSE=False

def load_models_OG(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,model_state,model_name,seed,temp,tokens,top_p,rep_p,chat_mem,cust_p):
    #token max=8192
    model_n=models[model_name]
    print(model_state)
    hist_len=0
    client=model_state
    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_choose(f"{system_prompt}, {prompt}", memory[0-chat_mem:],cust_p,model_name)
        else:
            formatted_prompt = format_prompt_choose(prompt, memory[0-chat_mem:],cust_p,model_name)
        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))        

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 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:
    model_state=gr.State()
    memory=gr.State()
    gr.HTML("""<center><h1 style='font-size:xx-large;'>LangHub</h1><br><h3>Fast Inference Playground</h3></center>""")
    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")                
                        test_btn=gr.Button("Test")
                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)
        test_json=gr.JSON(label="Test Output")
    test_btn.click(test_models,None,test_json)
    
    client_choice.change(load_models,client_choice,[chat_b,custom_prompt,model_state])
    app.load(load_models,client_choice,[chat_b,custom_prompt,model_state])
    
    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,model_state,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,model_state,client_choice,seed,temp,tokens,top_p,rep_p,chat_mem,custom_prompt],[chat_b,memory])
    
    stop_btn.click(None,None,None,cancels=[go,im_go,chat_sub])
    clear_btn.click(clear_fn,None,[inp,sys_inp,chat_b,memory])
app.queue(default_concurrency_limit=10).launch()