Spaces:
Sleeping
Sleeping
File size: 4,385 Bytes
df293a9 967880e 550c0fc 967880e 409b443 967880e 3df1aa2 f4b9b87 967880e 409b443 7288e51 2230706 967880e df293a9 f4b9b87 df293a9 7894e28 967880e 409b443 967880e 6274930 967880e df293a9 7288e51 df293a9 7288e51 df293a9 |
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 |
import os
os.system("pip uninstall -y gradio")
os.system("pip install gradio==3.50.2")
from huggingface_hub import InferenceClient
import gradio as gr
"""
Chat engine.
TODOs:
- Better prompts.
- Output reader / parser.
- Agents for evaluation and task planning / splitting.
* Haystack for orchestration
- Tools for agents
* Haystack for orchestration
-
"""
selected_model = "mistralai/Mixtral-8x7B-Instruct-v0.1"
client = InferenceClient(selected_model)
def format_prompt(query, history, lookback):
prompt = "Responses should be no more than 100 words long.\n"
for previous_query, prevous_completion in history[-lookback:]:
prompt += f"<s>[INST] {previous_query} [/INST] {prevous_completion}</s> "
prompt += f"[INST] {query} [/INST]"
return prompt
def query_submit(user_message, history):
return "", history + [[user_message, None]]
def query_completion(
query,
history,
lookback = 3,
max_new_tokens = 256,
):
generateKwargs = dict(
max_new_tokens = max_new_tokens,
seed = 1337,
)
formatted_query = format_prompt(query, history, lookback)
stream = client.text_generation(
formatted_query,
**generateKwargs,
stream = True,
details = True,
return_full_text = False
)
history[-1][1] = ""
for response in stream:
history[-1][1] += response.token.text
yield history
def retry_query(
history,
lookback = 3,
max_new_tokens = 256,
):
if not history:
pass
else:
query = history[-1][0]
history[-1][1] = None
generateKwargs = dict(
max_new_tokens = max_new_tokens,
seed = 1337,
)
formatted_query = format_prompt(query, history, lookback)
stream = client.text_generation(
formatted_query,
**generateKwargs,
stream = True,
details = True,
return_full_text = False
)
history[-1][1] = ""
for response in stream:
history[-1][1] += response.token.text
yield history
"""
Chat UI using Gradio Blocks.
"""
with gr.Blocks() as chatUI:
# gr.State()
with gr.Row():
with gr.Column(scale=1, min_width=200):
gr.Markdown(
r"Query History"
)
with gr.Column(scale=4):
gr.Textbox(
placeholder = "Please enter you question or request here...",
)
with gr.Group():
with gr.Row():
gr.Textbox(
placeholder = "Please enter you question or request here...",
show_label = False,
)
gr.Button("Submit")
gr.Chatbot(
bubble_full_width = False,
)
gr.Textbox(
placeholder = "Please enter you question or request here...",
show_label = False,
)
gr.Button("Submit")
with gr.Column(scale=3):
companyInfo = gr.Markdown(
r"Company Info"
)
companyBull = gr.Markdown(
r"Bull Case"
)
companyBear = gr.Markdown(
r"Bear Case"
)
"""
Event functions
"""
# queryInput.submit(
# fn = query_submit,
# inputs = [queryInput, chatOutput],
# outputs = [queryInput, chatOutput],
# queue = False,
# ).then(
# fn = query_completion,
# inputs = [queryInput, chatOutput],
# outputs = [chatOutput],
# )
# submitButton.click(
# fn = query_submit,
# inputs = [queryInput, chatOutput],
# outputs = [queryInput, chatOutput],
# queue = False,
# ).then(
# fn = query_completion,
# inputs = [queryInput, chatOutput],
# outputs = [chatOutput],
# )
# retryButton.click(
# fn = retry_query,
# inputs = [chatOutput],
# outputs = [chatOutput],
# )
chatUI.queue()
chatUI.launch(show_api = False) |