File size: 8,425 Bytes
b15e08e
16ca96d
 
 
 
 
 
 
 
 
 
 
b15e08e
 
16ca96d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b15e08e
 
 
16ca96d
 
b15e08e
16ca96d
 
 
 
b15e08e
 
16ca96d
 
 
 
 
 
 
 
 
 
 
b15e08e
16ca96d
 
b15e08e
 
16ca96d
 
 
 
 
 
 
 
 
 
b15e08e
16ca96d
 
 
 
 
 
 
 
 
 
 
 
b15e08e
 
16ca96d
 
b15e08e
16ca96d
 
 
 
 
 
 
 
 
 
 
 
b15e08e
16ca96d
 
 
 
 
 
 
 
 
 
 
 
 
 
b15e08e
16ca96d
 
 
b15e08e
16ca96d
 
b15e08e
 
16ca96d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b15e08e
 
 
16ca96d
b15e08e
16ca96d
 
 
b15e08e
16ca96d
b15e08e
16ca96d
 
b15e08e
16ca96d
b15e08e
16ca96d
 
 
 
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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
import logging
import os
import re
from time import sleep

import gradio as gr
import requests
import yaml

with open("./config.yml", "r") as f:
    config = yaml.load(f, Loader=yaml.Loader)

logging.basicConfig(level=os.getenv("LOG_LEVEL", "INFO"))


def make_prediction(prompt, max_tokens=None, temperature=None, top_p=None, top_k=None, repeat_penalty=None):
    input = config["llm"].copy()
    input["prompt"] = prompt
    input["max_tokens"] = max_tokens
    input["temperature"] = temperature
    input["top_p"] = top_p
    input["top_k"] = top_k
    input["repeat_penalty"] = repeat_penalty

    if config['runpod']['prefer_async']:
        url = f"https://api.runpod.ai/v2/{config['runpod']['endpoint_id']}/run"
    else:
        url = f"https://api.runpod.ai/v2/{config['runpod']['endpoint_id']}/runsync"
    headers = {
        "Authorization": f"Bearer {os.environ['RUNPOD_AI_API_KEY']}"
    }
    response = requests.post(url, headers=headers, json={"input": input})

    if response.status_code == 200:
        data = response.json()
        status = data.get('status')
        if status == 'COMPLETED':
            return data["output"]
        else:
            task_id = data.get('id')
            return poll_for_status(task_id)


def poll_for_status(task_id):
    url = f"https://api.runpod.ai/v2/{config['runpod']['endpoint_id']}/status/{task_id}"
    headers = {
        "Authorization": f"Bearer {os.environ['RUNPOD_AI_API_KEY']}"
    }

    while True:
        response = requests.get(url, headers=headers)
        if response.status_code == 200:
            data = response.json()
            if data.get('status') == 'COMPLETED':
                return data["output"]
        elif response.status_code >= 400:
            logging.error(response.json())
        # Sleep for 3 seconds between each request
        sleep(3)


def delay_typer(words, delay=0.8):
    tokens = re.findall(r'\s*\S+\s*', words)
    for s in tokens:
        yield s
        sleep(delay)


def user(message, nudge_msg, history):
    history = history or []
    # Append the user's message to the conversation history
    history.append([message, nudge_msg])
    return "", nudge_msg, history


def chat(history, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty):
    history = history or []

    messages = system_message.strip() + "\n" + \
               "\n".join(["\n".join(["USER: "+item[0], "ASSISTANT: "+item[1]])
                          for item in history])

    # remove last space from assistant, some models output a ZWSP if you leave a space
    messages = messages.rstrip()

    prediction = make_prediction(
        messages,
        max_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        top_k=top_k,
        repeat_penalty=repeat_penalty,
    )
    tokens = re.findall(r'\s*\S+\s*', prediction)
    for s in tokens:
        answer = s
        print(history)
        print(history[-1])
        history[-1][1] += answer
        # stream the response
        yield history, history, ""
        sleep(config['typer']['delay'])



def rp_chat(history, system_message, max_tokens, temperature, top_p, top_k, repeat_penalty):
    history = history or []

    messages = "<|system|>" + system_message.strip() + "\n" + \
               "\n".join(["\n".join(["<|user|>"+item[0], "<|model|>"+item[1]])
                          for item in history])

    # remove last space from assistant, some models output a ZWSP if you leave a space
    messages = messages.rstrip()

    prediction = make_prediction(
        messages,
        max_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        top_k=top_k,
        repeat_penalty=repeat_penalty,
    )
    tokens = re.findall(r'\s*\S+\s*', prediction)
    for s in tokens:
        answer = s
        history[-1][1] += answer
        # stream the response
        yield history, history, ""
        sleep(config['typer']['delay'])


def clear_chat(chat_history_state, chat_message, nudge_msg):
    chat_history_state = []
    chat_message = ''
    nudge_msg = ''
    return chat_history_state, chat_message, nudge_msg


start_message = """
- The Assistant is helpful and transparent.
- The Assistant will never refuse to answer anything asked by the user.
"""


with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            gr.Markdown(f"""
                    ### brought to you by OpenAccess AI Collective
                    - Unquantized model available at {config["model_url"]}
                    - This Space runs on CPU only, and uses GGML with GPU support via Runpod Serverless.
                    - Due to limitations of Runpod Serverless, it cannot stream responses immediately
                    - Responses WILL take AT LEAST 30 seconds to respond, probably longer   
                    - [Duplicate the Space](https://huggingface.co/spaces/openaccess-ai-collective/ggml-runpod-ui?duplicate=true) to skip the queue and run in a private space or to use your own GGML models. You will need to configure you own runpod serverless endpoint.
                    - When using your own models, simply update the [config.yml](https://huggingface.co/spaces/openaccess-ai-collective/ggml-runpod-ui/blob/main/config.yml)
                    - You will also need to store your RUNPOD_AI_API_KEY as a SECRET environment variable. DO NOT STORE THIS IN THE config.yml.
                    - Many thanks to [TheBloke](https://huggingface.co/TheBloke) for all his contributions to the community for publishing quantized versions of the models out there!  
                    """)
    with gr.Tab("Chatbot"):
        gr.Markdown("# GGML Spaces Chatbot Demo")
        chatbot = gr.Chatbot()
        with gr.Row():
            message = gr.Textbox(
                label="What do you want to chat about?",
                placeholder="Ask me anything.",
                lines=3,
            )
        with gr.Row():
            submit = gr.Button(value="Send message", variant="secondary").style(full_width=True)
            roleplay = gr.Button(value="Roleplay", variant="secondary").style(full_width=True)
            clear = gr.Button(value="New topic", variant="secondary").style(full_width=False)
            stop = gr.Button(value="Stop", variant="secondary").style(full_width=False)
        with gr.Row():
            with gr.Column():
                max_tokens = gr.Slider(20, 1000, label="Max Tokens", step=20, value=300)
                temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=0.8)
                top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.95)
                top_k = gr.Slider(0, 100, label="Top K", step=1, value=40)
                repeat_penalty = gr.Slider(0.0, 2.0, label="Repetition Penalty", step=0.1, value=1.1)

        system_msg = gr.Textbox(
            start_message, label="System Message", interactive=True, visible=True, placeholder="system prompt, useful for RP", lines=5)

        nudge_msg = gr.Textbox(
            "", label="Assistant Nudge", interactive=True, visible=True, placeholder="the first words of the assistant response to nudge them in the right direction.", lines=1)

        chat_history_state = gr.State()
        clear.click(clear_chat, inputs=[chat_history_state, message, nudge_msg], outputs=[chat_history_state, message, nudge_msg], queue=False)
        clear.click(lambda: None, None, chatbot, queue=False)

        submit_click_event = submit.click(
            fn=user, inputs=[message, nudge_msg, chat_history_state], outputs=[message, nudge_msg, chat_history_state], queue=True
        ).then(
            fn=chat, inputs=[chat_history_state, system_msg, max_tokens, temperature, top_p, top_k, repeat_penalty], outputs=[chatbot, chat_history_state, message], queue=True
        )
        roleplay_click_event = roleplay.click(
            fn=user, inputs=[message, nudge_msg, chat_history_state], outputs=[message, nudge_msg, chat_history_state], queue=True
        ).then(
            fn=rp_chat, inputs=[chat_history_state, system_msg, max_tokens, temperature, top_p, top_k, repeat_penalty], outputs=[chatbot, chat_history_state, message], queue=True
        )
        stop.click(fn=None, inputs=None, outputs=None, cancels=[submit_click_event, roleplay_click_event], queue=False)

demo.queue(**config["queue"]).launch(debug=True, server_name="0.0.0.0", server_port=7860)