Spaces:
Running
Running
File size: 4,329 Bytes
ac9e7e2 43ba4e0 ac9e7e2 43ba4e0 342fff9 43ba4e0 ac9e7e2 8bff249 43ba4e0 ac9e7e2 43ba4e0 ac9e7e2 |
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 |
import requests
import json
import streamlit as st
import sys
import time
st.set_page_config(page_title="Samba-CoE_v0.1 Chatbot")
if 'expert_sampling_details' not in st.session_state.keys():
st.session_state['temperature'] = 0.7
st.session_state['top_p'] = 0.1
st.session_state['max_length'] = 512
st.session_state['history_buffer_length'] = 3
st.session_state['top_k'] = 40
with st.sidebar:
st.sidebar.info("Samba-CoE_v0.1 Chatbot")
with st.sidebar.form("Chatbot Settings"):
temperature = st.sidebar.slider('temperature', min_value=0.01, max_value=1.0, value=0.7, step=0.01)
top_p = st.sidebar.slider('top_p', min_value=0.01, max_value=1.0, value=0.1, step=0.01)
max_length = st.sidebar.slider('max_length', min_value=32, max_value=1024, value=512, step=8)
submitted = st.form_submit_button("Submit")
if submitted:
st.session_state['max_length'] = max_length
st.session_state['temperature'] = temperature
st.session_state['top_p'] = top_p
st.sidebar.success("Generation parameters:\n" + f'\n * temperature: {st.session_state.temperature}\n' + f'\n * top_p: {st.session_state.top_p}\n' + f'\n * max_tokens: {st.session_state.max_length}\n' + f'\n * skip_special_tokens: True\n'+ f'\n * do_sample: True\n' )
def clear_chat_history():
st.session_state.messages = [{"role": "assistant", "content": "How may I assist you today?"}]
st.sidebar.button('Clear Chat History', on_click=clear_chat_history)
# Store LLM generated responses
if "messages" not in st.session_state.keys():
st.session_state.messages = [{"role": "assistant", "content": "How may I assist you today?"}]
# Display or clear chat message
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.write(message["content"])
def generate_response(prompt_input):
history_buffer_length = st.session_state.history_buffer_length
message_history = st.session_state.messages[-history_buffer_length:]
string_dialogue = "<s> [INST] You are a helpful assistant developed by SambaNova Systems as part of its Composition of Expert (CoE) effort. Always assist with care, respect, and truth. Respond with utmost utility yet securely and professionally. Avoid harmful, unethical, prejudiced, or negative content. Ensure replies promote fairness, positivity and an engaging conversation. [/INST] \n"
for dict_message in message_history:
if dict_message["role"] == "user":
string_dialogue += '[INST]' + dict_message["content"] + '[/INST]' + '\n'
else:
string_dialogue += dict_message["content"] + "</s>" + '\n'
payload = {'prompt': string_dialogue,
'max_tokens': st.session_state['max_length'],
'n': 1,
'do_sample': True,
'temperature': st.session_state['temperature'],
'top_p': st.session_state['top_p'],
'top_k': st.session_state['top_k'],
'skip_special_token': True,
'repition_penalty': 1.15,
'stop_sequences': ['INST', '[INST]', '[/INST]']
}
r = requests.post(st.secrets["backend_url"], json=payload)
response = r.json()['choices'][0]['text']
return response
# User-provided prompt
if prompt := st.chat_input():
st.session_state.messages.append({"role": "user", "content": prompt})
with st.chat_message("user"):
st.write(prompt)
def clean_response(response):
if '[INST' in response:
return response.split('[INST')[0]
elif '[/INST' in response:
return response.split('[/INST')[0]
return response
def stream_response(response):
response = "Please access our latest chatbot at this link: [Samba-CoE-v0.2](https://coe-1.cloud.snova.ai/)"
for word in response.split(" "):
yield word + " "
time.sleep(0.02)
# Generate a new response if last message is not from assistant
if st.session_state.messages[-1]["role"] != "assistant":
with st.chat_message("assistant"):
with st.spinner("Thinking..."):
# response = generate_response(prompt)
# response = clean_response(response)
response = None
st.write_stream(stream_response(response))
message = {"role": "assistant", "content": response}
st.session_state.messages.append(message)
|