File size: 2,196 Bytes
77ea5bf
 
ffb6bec
 
 
 
77ea5bf
a9a712b
 
 
 
 
77ea5bf
ffb6bec
 
 
 
77ea5bf
ffb6bec
 
 
 
 
77ea5bf
 
ffb6bec
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
77ea5bf
ffb6bec
 
 
 
 
77ea5bf
 
 
ffb6bec
 
 
 
 
77ea5bf
ffb6bec
 
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
57
58
59
60
61
62
63
64
65
66
import gradio as gr
from huggingface_hub import InferenceClient
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
import os
from threading import Thread


from datasets import load_dataset

ds = load_dataset("hichri-mo/twensa-hositng")
context=ds["text"]

system_prompt = f"""you are twensa hosting chat bot
to know more about twensa hosting this is an ad about them :
{context}
"""

model = AutoModelForCausalLM.from_pretrained("KingNish/Qwen2.5-0.5b-Test-ft",
                                             torch_dtype=torch.float16)
tokenizer = AutoTokenizer.from_pretrained("KingNish/Qwen2.5-0.5b-Test-ft")
device = torch.device('cuda')
model = model.to(device)


def chat(message, history):
    chat = [{"role":"system","content":system_prompt}]
    for item in history:
        chat.append({"role": "user", "content": item[0]})
        if item[1] is not None:
            chat.append({"role": "assistant", "content": item[1]})
    chat.append({"role": "user", "content": message})
    messages = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
    # Tokenize the messages string
    model_inputs = tokenizer([messages], return_tensors="pt").to(device)
    streamer = TextIteratorStreamer(
        tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
    generate_kwargs = dict(
        model_inputs,
        streamer=streamer,
        max_new_tokens=1024,
        do_sample=True,
        top_p=0.95,
        top_k=1000,
        temperature=0.75,
        num_beams=1,
    )
    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    # Initialize an empty string to store the generated text
    partial_text = ""
    for new_text in streamer:
        partial_text += new_text
        yield partial_text



demo = gr.ChatInterface(fn=chat,
                        chatbot=gr.Chatbot(show_label=True, show_share_button=True, show_copy_button=True,layout="bubble", bubble_full_width=False),
                        theme="dark",
                        examples=[["what is twensa hosting ?"]],
                        title="TWENSA HOSTING CHAT BOT")

# Launch the app
demo.launch()