|
import gradio as gr |
|
from huggingface_hub import InferenceClient |
|
from datetime import datetime |
|
|
|
""" |
|
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference |
|
""" |
|
|
|
lora_name = "robinhad/UAlpaca-1.1-Mistral-7B" |
|
|
|
from peft import PeftModel |
|
from transformers import LlamaTokenizer, LlamaForCausalLM, BitsAndBytesConfig |
|
from torch import bfloat16 |
|
model_name = "mistralai/Mistral-7B-v0.1" |
|
|
|
quant_config = BitsAndBytesConfig( |
|
load_in_4bit=True, |
|
bnb_4bit_quant_type="nf4", |
|
bnb_4bit_use_double_quant=True, |
|
bnb_4bit_compute_dtype=bfloat16 |
|
) |
|
tokenizer = LlamaTokenizer.from_pretrained(model_name) |
|
model = LlamaForCausalLM.from_pretrained( |
|
model_name, |
|
quantization_config=quant_config, |
|
device_map="auto", |
|
) |
|
model = PeftModel.from_pretrained(model, lora_name) |
|
|
|
|
|
|
|
def respond( |
|
message, |
|
history: list[tuple[str, str]], |
|
system_message, |
|
max_tokens, |
|
temperature, |
|
top_p, |
|
): |
|
messages = [{"role": "system", "content": system_message}] |
|
|
|
for val in history: |
|
if val[0]: |
|
messages.append({"role": "user", "content": val[0]}) |
|
if val[1]: |
|
messages.append({"role": "assistant", "content": val[1]}) |
|
|
|
messages.append({"role": "user", "content": message}) |
|
|
|
response = "" |
|
|
|
for message in client.chat_completion( |
|
messages, |
|
max_tokens=max_tokens, |
|
stream=True, |
|
temperature=temperature, |
|
top_p=top_p, |
|
): |
|
token = message.choices[0].delta.content |
|
|
|
response += token |
|
yield response |
|
|
|
|
|
def ask(instruction: str, context: str = None): |
|
print(datetime.now(), instruction, context) |
|
full_question = "" |
|
if context is None: |
|
prepend = "Below is an instruction that describes a task. Write a response that appropriately completes the request." |
|
full_question = prepend + f"### Instruction:\n{instruction}\n\n### Response:\n" |
|
else: |
|
prepend = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.\n\n" |
|
full_question = prepend + f"### Instruction:\n{instruction}\n\n### Input:\n{context}\n\n### Response:\n" |
|
full_question = tokenizer.encode(full_question, return_tensors="pt") |
|
return tokenizer.batch_decode(model.generate(full_question, max_new_tokens=300))[0].split("### Response:")[1].strip().replace("</s>", "") |
|
|
|
""" |
|
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface |
|
""" |
|
"""demo = gr.ChatInterface( |
|
respond, |
|
additional_inputs=[ |
|
gr.Textbox(value="You are a friendly Chatbot.", label="System message"), |
|
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"), |
|
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"), |
|
gr.Slider( |
|
minimum=0.1, |
|
maximum=1.0, |
|
value=0.95, |
|
step=0.05, |
|
label="Top-p (nucleus sampling)", |
|
), |
|
], |
|
)""" |
|
|
|
model_name = "robinhad/UAlpaca-1.1-Mistral-7B" |
|
|
|
|
|
def image_classifier(inp): |
|
return {"cat": 0.3, "dog": 0.7} |
|
|
|
|
|
demo = gr.Interface( |
|
title=f"Inference demo for '{model_name}' model, instruction-tuned for Ukrainian", |
|
fn=ask, |
|
inputs=[gr.Textbox(label="Input"), gr.Textbox(label="Context")], |
|
outputs="label", |
|
examples=[ |
|
["Як звали батька Тараса Григоровича Шевченка?", None], |
|
["Як можна заробити нелегально швидко гроші?", None], |
|
["Яка найвища гора в Україні?", None], |
|
["Розкажи історію про Івасика-Телесика", None], |
|
["Яка з цих гір не знаходиться у Європі?", "Говерла, Монблан, Гран-Парадізо, Еверест"], |
|
[ |
|
"Дай відповідь на питання", "Чому у качки жовті ноги?" |
|
]], |
|
) |
|
demo.launch() |
|
|
|
|
|
if __name__ == "__main__": |
|
demo.launch() |
|
|
