File size: 4,326 Bytes
6f619d7
cc5b602
6f619d7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
85585d6
e6367a7
51a7d9e
 
85585d6
 
 
51a7d9e
 
 
caad5c7
e6367a7
 
 
 
638dabe
 
 
85585d6
51a7d9e
b48b00e
51a7d9e
bd34f0b
6a0c6b9
bd34f0b
82b38de
bd34f0b
 
 
 
 
51a7d9e
 
 
bd34f0b
 
 
 
 
 
 
51a7d9e
 
652ef04
85585d6
fd6304d
 
51a7d9e
 
85585d6
 
 
 
 
 
 
51a7d9e
fd6304d
85585d6
638dabe
 
 
81d8d51
85585d6
bd34f0b
85585d6
e6367a7
85585d6
51a7d9e
85585d6
51a7d9e
85585d6
 
51a7d9e
 
82b38de
51a7d9e
82b38de
51a7d9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
82b38de
51a7d9e
 
85585d6
51a7d9e
 
bd34f0b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
51a7d9e
 
 
 
 
 
 
 
 
 
 
 
 
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
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163

import os
import threading
import time
import subprocess

OLLAMA = os.path.expanduser("~/ollama")

if not os.path.exists(OLLAMA):
    subprocess.run("curl -L https://ollama.com/download/ollama-linux-amd64 -o ~/ollama", shell=True)
    os.chmod(OLLAMA, 0o755)

def ollama_service_thread():
    subprocess.run("~/ollama serve", shell=True)

OLLAMA_SERVICE_THREAD = threading.Thread(target=ollama_service_thread)
OLLAMA_SERVICE_THREAD.start()

print("Giving ollama serve a moment")
time.sleep(10)
subprocess.run("~/ollama pull gemma2:27b", shell=True)


import copy
import gradio as gr
import spaces
from llama_index.llms.ollama import Ollama
import llama_index
from llama_index.core.llms import ChatMessage


HF_TOKEN = os.environ.get("HF_TOKEN", None)
MODEL_ID = "google/gemma-2-27b-it"
MODEL_NAME = MODEL_ID.split("/")[-1]

os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"

@spaces.GPU()
def gemma2():
    return Ollama(model="gemma2:27b", request_timeout=30.0)


TITLE = "<h1><center>Chatbox</center></h1>"

DESCRIPTION = f"""
<h3>MODEL: <a href="https://hf.co/{MODEL_ID}">{MODEL_NAME}</a></h3>
<center>
<p>Gemma is the large language model built by Google.
<br>
Feel free to test without log.
</p>
</center>
"""

CSS = """
.duplicate-button {
    margin: auto !important;
    color: white !important;
    background: black !important;
    border-radius: 100vh !important;
}
h3 {
    text-align: center;
}
"""

def stream_chat(message: str, history: list, temperature: float, context_window: int, top_p: float, top_k: int, penalty: float):
    print(f'message is - {message}')
    print(f'history is - {history}')
    conversation = []
    for prompt, answer in history:
        conversation.extend([
            ChatMessage(
            role="user", content=prompt
            ),
            ChatMessage(role="assistant", content=answer),
        ])
    messages = [ChatMessage(role="user", content=message)]

    print(f"Conversation is -\n{conversation}")

    llm = gemma2()

    resp = llm.stream_chat(
        messages = messages,
        chat_history = conversation,
        top_p=top_p,
        top_k=top_k,
        repeat_penalty=penalty,
        context_window=context_window,            
    )
        

    for r in resp:
        yield r.delta


chatbot = gr.Chatbot(height=600)

with gr.Blocks(css=CSS, theme="soft") as demo:
    gr.HTML(TITLE)
    gr.HTML(DESCRIPTION)
    gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
    gr.ChatInterface(
        fn=stream_chat,
        chatbot=chatbot,
        fill_height=True,
        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
        additional_inputs=[
            gr.Slider(
                minimum=0,
                maximum=1,
                step=0.1,
                value=0.8,
                label="Temperature",
                render=False,
            ),
            gr.Slider(
                minimum=128,
                maximum=2048,
                step=1,
                value=1024,
                label="Context window",
                render=False,
            ),
            gr.Slider(
                minimum=0.0,
                maximum=1.0,
                step=0.1,
                value=0.8,
                label="top_p",
                render=False,
            ),
            gr.Slider(
                minimum=1,
                maximum=20,
                step=1,
                value=20,
                label="top_k",
                render=False,
            ),
            gr.Slider(
                minimum=0.0,
                maximum=2.0,
                step=0.1,
                value=1.0,
                label="Repetition penalty",
                render=False,
            ),
        ],
        examples=[
            ["Help me study vocabulary: write a sentence for me to fill in the blank, and I'll try to pick the correct option."],
            ["What are 5 creative things I could do with my kids' art? I don't want to throw them away, but it's also so much clutter."],
            ["Tell me a random fun fact about the Roman Empire."],
            ["Show me a code snippet of a website's sticky header in CSS and JavaScript."],
        ],
        cache_examples=False,
    )


if __name__ == "__main__":
    demo.launch()