Spaces:
Runtime error
Runtime error
import gradio as gr | |
from gradio_client import Client | |
from huggingface_hub import InferenceClient | |
import random | |
ss_client = Client("https://xilixmeaty40-html-image-current-tabx.hf.space/") | |
with open("models.txt", "r") as file: | |
models = file.read().splitlines() | |
combined_model = "\n\n".join(models) | |
try: | |
client = InferenceClient(combined_model) | |
except Exception as e: | |
raise Exception(f"Failed to load models: {e}") | |
def load_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>" | |
prompt += cust_p.replace("USER_INPUT", message) | |
return prompt | |
def chat_inf(system_prompt, prompt, history, memory, seed, temp, tokens, top_p, rep_p, chat_mem, cust_p): | |
hist_len = 0 | |
if not history: | |
history = [] | |
if not memory: | |
memory = [] | |
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) | |
try: | |
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True, timeout=10) | |
output = "" | |
for response in stream: | |
output += response.token.text | |
yield [(prompt, output)], memory | |
history.append((prompt, output)) | |
memory.append((prompt, output)) | |
yield history, memory | |
except Exception as e: | |
print(f"Error during model inference: {e}") | |
yield [("Error", "The model failed to respond, possibly due to a timeout. Please try again.")], memory | |
def get_screenshot(chat, height=5000, width=600, chatblock=[], theme="light", wait=3000, header=True): | |
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://xilixmeaty40-html-image-current-tabx.hf.space/file={result[tog]}' | |
return out | |
def clear_fn(): | |
return None, None, None, None | |
rand_val = random.randint(1, 1111111111111111) | |
def check_rand(inp, val): | |
if inp: | |
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() | |
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)") | |
custom_prompt = gr.Textbox(label="Modify Prompt Format", 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): | |
stop_btn = gr.Button("Stop") | |
clear_btn = gr.Button("Clear") | |
seed = gr.Slider(label="Seed", minimum=1, maximum=1111111111111111, step=1, value=rand_val) | |
tokens = gr.Slider(label="Max new tokens", value=300000, minimum=0, maximum=800000, step=64) | |
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", value=4) | |
with gr.Accordion(label="Screenshot", open=False): | |
im_btn = gr.Button("Screenshot") | |
img = gr.Image(type='filepath') | |
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", choices=list(range(0, 21)), value=0, type="index") | |
header = gr.Checkbox(label="Include header?", value=True) | |
check_rand(rand_val, rand_val) | |
btn.click(chat_inf, inputs=[sys_inp, inp, chat_b, memory, seed, temp, tokens, top_p, rep_p, chat_mem, custom_prompt], outputs=[chat_b, memory]) | |
stop_btn.click(lambda: None, []) | |
clear_btn.click(clear_fn, []) | |
app.launch(share=True) | |