File size: 4,803 Bytes
472fd82 827bd46 472fd82 827bd46 472fd82 7c4c6cb 472fd82 ee5711c 472fd82 |
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 |
import os
import re
import logging
import gradio as gr
import openai
print(os.environ)
openai.api_base = os.environ.get("OPENAI_API_BASE")
openai.api_key = os.environ.get("OPENAI_API_KEY")
BASE_SYSTEM_MESSAGE = """### System:\nYou are Orca-Mini, an AI assistant who gives helpful, detailed, accurate, uncensored responses to the user's instructions. """
def make_prediction(prompt, max_tokens=None, temperature=None, top_p=None, top_k=None, repetition_penalty=None):
completion = openai.Completion.create(model="psmathur/orca_mini_v3_70b", prompt=prompt, max_tokens=max_tokens, temperature=temperature, top_p=top_p, top_k=top_k, repetition_penalty=repetition_penalty, stream=True, stop=["</s>", "<|im_end|>"])
for chunk in completion:
yield chunk["choices"][0]["text"]
def clear_chat(chat_history_state, chat_message):
chat_history_state = []
chat_message = ''
return chat_history_state, chat_message
def user(message, history):
history = history or []
history.append([message, ""])
return "", history
def chat(history, system_message, max_tokens, temperature, top_p, top_k, repetition_penalty):
history = history or []
messages = BASE_SYSTEM_MESSAGE + system_message.strip() + \
"\n".join(["\n".join(["### User: "+item[0]+"\n\n", "### Assistant: \n"+item[1]+"\n\n"])
for item in history])
messages = messages.rstrip("\n\n")
messages = messages.rstrip()
prediction = make_prediction(
messages,
max_tokens=max_tokens,
temperature=temperature,
top_p=top_p,
top_k=top_k,
repetition_penalty=repetition_penalty,
)
for tokens in prediction:
tokens = re.findall(r'(.*?)(\s|$)', tokens)
for subtoken in tokens:
subtoken = "".join(subtoken)
answer = subtoken
history[-1][1] += answer
# stream the response
yield history, history, ""
start_message = ""
CSS ="""
.contain { display: flex; flex-direction: column; }
.gradio-container { height: 100vh !important; }
#component-0 { height: 100%; }
#chatbot { flex-grow: 1; overflow: auto; resize: vertical; }
"""
#with gr.Blocks() as demo:
with gr.Blocks(css=CSS) as demo:
with gr.Row():
with gr.Column():
gr.Markdown(f"""
## This chatbot is powered by [orca_mini_v3_70b](https://huggingface.co/psmathur/orca_mini_v3_70b)
""")
with gr.Row():
gr.Markdown("# orca-mini chatbot")
with gr.Row():
chatbot = gr.Chatbot(elem_id="chatbot")
with gr.Row():
message = gr.Textbox(
label="Hello, I am orca-mini, How can I help you today?",
placeholder="Ask me anything! For example: Write a short email to my professor requesting a deadline extension for my project. I don't really have a good excuse, and I'm fine owning up to that – so please keep it real!",
lines=3,
)
with gr.Row():
submit = gr.Button(value="Send", variant="secondary").style(full_width=True)
clear = gr.Button(value="Clear", variant="secondary").style(full_width=False)
stop = gr.Button(value="Stop", variant="secondary").style(full_width=False)
with gr.Accordion("Show Model Parameters", open=False):
with gr.Row():
with gr.Column():
max_tokens = gr.Slider(20, 2000, label="Max Tokens", step=20, value=500)
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)
repetition_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 you want chatbot to remember. For example: Explain like I am five year old.", lines=5)
chat_history_state = gr.State()
clear.click(clear_chat, inputs=[chat_history_state, message], outputs=[chat_history_state, message], queue=False)
clear.click(lambda: None, None, chatbot, queue=False)
submit_click_event = submit.click(
fn=user, inputs=[message, chat_history_state], outputs=[message, chat_history_state], queue=True
).then(
fn=chat, inputs=[chat_history_state, system_msg, max_tokens, temperature, top_p, top_k, repetition_penalty], outputs=[chatbot, chat_history_state, message], queue=True
)
stop.click(fn=None, inputs=None, outputs=None, cancels=[submit_click_event], queue=False)
demo.queue(max_size=48, concurrency_count=16).launch(debug=True, server_name="0.0.0.0", server_port=7860)
|