File size: 2,005 Bytes
8e3ee33
4ab0120
8e3ee33
 
 
4b787b5
96ef64b
8e3ee33
6c49eee
4b787b5
 
 
 
 
 
 
 
 
 
 
 
 
8e3ee33
 
 
 
 
 
 
 
 
4b787b5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ab0120
c0799ff
4b787b5
4ab0120
8e3ee33
6c49eee
36a2111
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
import os
os.system("pip install ctransformers gradio")
import time
import requests
from tqdm import tqdm
import ctransformers
import gradio as gr

if not os.path.isfile('./llama-2-7b.ggmlv3.q4_K_S.bin'):
    print("Downloading Model from HuggingFace")
    url = "https://huggingface.co/TheBloke/Llama-2-7B-GGML/resolve/main/llama-2-7b.ggmlv3.q4_K_S.bin"
    response = requests.get(url, stream=True)
    total_size_in_bytes = int(response.headers.get('content-length', 0))
    block_size = 1024  # 1 Kibibyte
    progress_bar = tqdm(total=total_size_in_bytes, unit='iB', unit_scale=True)
    with open('llama-2-7b.ggmlv3.q4_K_S.bin', 'wb') as file:
        for data in response.iter_content(block_size):
            progress_bar.update(len(data))
            file.write(data)
    progress_bar.close()
    if total_size_in_bytes != 0 and progress_bar.n != total_size_in_bytes:
        print("ERROR, something went wrong")

configObj = ctransformers.Config(stop=["\n", 'User'])
config = ctransformers.AutoConfig(config=configObj, model_type='llama')
config.config.stop = ["\n"]

llm = ctransformers.AutoModelForCausalLM.from_pretrained('./llama-2-7b.ggmlv3.q4_K_S.bin', config=config)
print("Loaded model")

def complete(prompt, stop=["User", "Assistant"]):
    tokens = llm.tokenize(prompt)
    token_count = 0
    output = ''
    for token in llm.generate(tokens):
        token_count += 1
        result = llm.detokenize(token)
        output += result
        for word in stop:
            if word in output:
                print('\n')
                return [output, token_count]
        print(result, end='', flush=True)

    print('\n')
    return [output, token_count]

def greet(question):
    print(question)
    output, token_count = complete(f'User: {question}. Can you please answer this as informatively but concisely as possible.\nAssistant: ')
    return f"Response: {output} | Tokens: {token_count}"

iface = gr.Interface(fn=greet, inputs="text", outputs="text")
iface.launch(share=True)