import gradio as gr
import os
from pathlib import Path
import argparse
from huggingface_hub import snapshot_download
from llama_cpp import Llama
repo_name = 'tolgadev/llama-2-7b-tk-mini-GGUF'
model_file = "llama-2-7b-tk-mini.Q3_K_M.gguf"
snapshot_download(repo_id=repo_name, local_dir=".", allow_patterns=model_file)
DEFAULT_MODEL_PATH = model_file
llm = Llama(model_path=model_file, model_type="llama")
def predict(input, chatbot, max_length, top_p, temperature, history):
chatbot.append((input, ""))
response = ""
history.append(input)
for output in llm(input, stream=True, temperature=temperature, top_p=top_p, max_tokens=max_length, ):
piece = output['choices'][0]['text']
response += piece
chatbot[-1] = (chatbot[-1][0], response)
yield chatbot, history
history.append(response)
yield chatbot, history
def reset_user_input():
return gr.update(value="")
def reset_state():
return [], []
with gr.Blocks() as demo:
gr.HTML("""
RuyaTabirleriLLM Chatbot Demo
This is unofficial demo of `tolgadev/llama-2-7b-tk-mini-GGUF` model based on LLama2 architecture.
Hit the like button if you liked! 🤗
""")
chatbot = gr.Chatbot()
with gr.Row():
with gr.Column(scale=4):
user_input = gr.Textbox(show_label=False, placeholder="Input...", lines=8, elem_id="user_input")
submitBtn = gr.Button("Submit", variant="primary", elem_id="submit_btn")
with gr.Column(scale=1):
max_length = gr.Slider(0, 2048, value=1024, step=2.0, label="Maximum Length", interactive=True)
top_p = gr.Slider(0, 1, value=0.7, step=0.01, label="Top P", interactive=True)
temperature = gr.Slider(0, 1.0, value=0.7, step=0.1, label="Temperature", interactive=True)
emptyBtn = gr.Button("Clear History")
history = gr.State([])
submitBtn.click(
predict, [user_input, chatbot, max_length, top_p, temperature, history], [chatbot, history], show_progress=True
)
submitBtn.click(reset_user_input, [], [user_input])
emptyBtn.click(reset_state, outputs=[chatbot, history], show_progress=True)
demo.queue().launch(share=False, inbrowser=True)