Spaces:
Running
Running
File size: 7,727 Bytes
a8bf50c b4cc323 a8bf50c b4cc323 a8bf50c 85ebe7f a8bf50c bc13def a8bf50c 6ca1259 a8bf50c c61084a a8bf50c 6ca1259 a8bf50c 424f4db a8bf50c |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 |
import os
import gradio as gr
from gradio.components import Textbox, Button, Slider, Checkbox
from AinaTheme import theme
from urllib.error import HTTPError
from rag import RAG
from utils import setup
MAX_NEW_TOKENS = 700
SHOW_MODEL_PARAMETERS_IN_UI = os.environ.get("SHOW_MODEL_PARAMETERS_IN_UI", default="True") == "True"
setup()
rag = RAG(
hf_token=os.getenv("HF_TOKEN"),
embeddings_model=os.getenv("EMBEDDINGS"),
repo_name=os.getenv("REPO_NAME"),
model_name=os.getenv("MODEL"),
)
# rerank_model=os.getenv("RERANK_MODEL"),
# rerank_number_contexts=int(os.getenv("RERANK_NUMBER_CONTEXTS"))
def generate(prompt, model_parameters):
try:
output, context, source = rag.get_response(prompt, model_parameters)
return output, context, source
except HTTPError as err:
if err.code == 400:
gr.Warning(
"The inference endpoint is only available Monday through Friday, from 08:00 to 20:00 CET."
)
except:
gr.Warning(
"Inference endpoint is not available right now. Please try again later."
)
return None, None, None
def submit_input(input_, num_chunks, max_new_tokens, repetition_penalty, top_k, top_p, do_sample, temperature):
if input_.strip() == "":
gr.Warning("Not possible to inference an empty input")
return None
model_parameters = {
"NUM_CHUNKS": num_chunks,
"max_new_tokens": max_new_tokens,
"repetition_penalty": repetition_penalty,
"top_k": top_k,
"top_p": top_p,
"do_sample": do_sample,
"temperature": temperature
}
output, context, source = generate(input_, model_parameters)
sources_markup = ""
for url in source:
sources_markup += f'<a href="{url}" target="_blank">{url}</a><br>'
return output, sources_markup, context
# return output.strip(), sources_markup, context
def change_interactive(text):
if len(text) == 0:
return gr.update(interactive=True), gr.update(interactive=False)
return gr.update(interactive=True), gr.update(interactive=True)
def clear():
return (
None,
None,
None,
None,
gr.Slider(value=2.0),
gr.Slider(value=MAX_NEW_TOKENS),
gr.Slider(value=1.0),
gr.Slider(value=50),
gr.Slider(value=0.99),
gr.Checkbox(value=False),
gr.Slider(value=0.35),
)
def gradio_app():
with gr.Blocks(theme=theme) as demo:
with gr.Row():
with gr.Column():
gr.Markdown(
"""#"""
)
with gr.Row(equal_height=True):
with gr.Column(variant="panel"):
input_ = Textbox(
lines=11,
label="Input",
placeholder=" ",
# value = "Quina és la finalitat del Servei Meteorològic de Catalunya?"
)
with gr.Row(variant="panel"):
clear_btn = Button(
"Clear",
)
submit_btn = Button("Submit", variant="primary", interactive=False)
with gr.Row(variant="panel"):
with gr.Accordion("Model parameters", open=False, visible=SHOW_MODEL_PARAMETERS_IN_UI):
num_chunks = Slider(
minimum=1,
maximum=6,
step=1,
value=2,
label="Number of chunks"
)
max_new_tokens = Slider(
minimum=50,
maximum=2000,
step=1,
value=MAX_NEW_TOKENS,
label="Max tokens"
)
repetition_penalty = Slider(
minimum=0.1,
maximum=2.0,
step=0.1,
value=1.0,
label="Repetition penalty"
)
top_k = Slider(
minimum=1,
maximum=100,
step=1,
value=50,
label="Top k"
)
top_p = Slider(
minimum=0.01,
maximum=0.99,
value=0.99,
label="Top p"
)
do_sample = Checkbox(
value=False,
label="Do sample"
)
temperature = Slider(
minimum=0.1,
maximum=1,
value=0.15,
label="Temperature"
)
parameters_compontents = [num_chunks, max_new_tokens, repetition_penalty, top_k, top_p, do_sample, temperature]
with gr.Column(variant="panel"):
output = Textbox(
lines=10,
label="Output",
interactive=False,
show_copy_button=True
)
with gr.Accordion("Sources and context:", open=False):
source_context = gr.Markdown(
label="Sources",
show_label=True,
)
with gr.Accordion("See full context evaluation:", open=False):
context_evaluation = gr.Markdown(
label="Full context",
show_label=False,
# interactive=False,
# autoscroll=False,
# show_copy_button=True
)
input_.change(
fn=change_interactive,
inputs=[input_],
outputs=[clear_btn, submit_btn],
api_name=False,
)
input_.change(
fn=None,
inputs=[input_],
api_name=False,
js="""(i, m) => {
document.getElementById('inputlenght').textContent = i.length + ' '
document.getElementById('inputlenght').style.color = (i.length > m) ? "#ef4444" : "";
}""",
)
clear_btn.click(
fn=clear,
inputs=[],
outputs=[input_, output, source_context, context_evaluation] + parameters_compontents,
queue=False,
api_name=False
)
submit_btn.click(
fn=submit_input,
inputs=[input_]+ parameters_compontents,
outputs=[output, source_context, context_evaluation],
api_name="get-results"
)
with gr.Row():
with gr.Column(scale=0.5):
gr.Examples(
examples=[
[""" How can i stratify patients with acute GVHD?"""], ["Which is the treament for grade 2 cytokine release syndrome?"],["What is the second line treatment for chronic GVHD?"]
],
inputs=input_,
outputs=[output, source_context, context_evaluation],
fn=submit_input,
)
demo.launch(show_api=True)
if __name__ == "__main__":
gradio_app() |