File size: 1,601 Bytes
c207c30 908daa6 c207c30 a9eba92 c207c30 908daa6 c207c30 1d65d17 c207c30 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 |
import gradio as gr
import random
import time
from ctransformers import AutoModelForCausalLM
from datetime import datetime
params = {
"max_new_tokens":512,
"stop":["<end>" ,"<|endoftext|>","[", "<user>"],
"temperature":0.7,
"top_p":0.8,
"stream":True,
"batch_size": 8}
llm = AutoModelForCausalLM.from_pretrained("Aspik101/trurl-2-7b-pl-instruct_GGML", model_type="llama")
with gr.Blocks() as demo:
chatbot = gr.Chatbot()
msg = gr.Textbox()
clear = gr.Button("Clear")
def user(user_message, history):
return "", history + [[user_message, None]]
def parse_history(hist):
history_ = ""
for q, a in hist:
history_ += f"<user>: {q } \n"
if a:
history_ += f"<assistant>: {a} \n"
return history_
def bot(history):
print(f"When: {datetime.today().strftime('%Y-%m-%d %H:%M:%S')}")
print("history: ",history)
prompt = f"Jesteś AI assystentem. Odpowiadaj po polsku. {parse_history(history)}. <assistant>:"
print("prompt: ",prompt)
stream = llm(prompt, **params)
history[-1][1] = ""
answer_save = ""
for character in stream:
history[-1][1] += character
answer_save += character
time.sleep(0.005)
yield history
print("answer_save: ",answer_save)
msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(
bot, chatbot, chatbot
)
clear.click(lambda: None, None, chatbot, queue=False)
demo.queue()
demo.launch() |