Spaces:
Runtime error
Runtime error
Commit
·
77ff0b3
1
Parent(s):
4bcccc4
Upload 2 files
Browse files- app.py +219 -0
- requirements.txt +14 -0
app.py
ADDED
|
@@ -0,0 +1,219 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# -*- coding:utf-8 -*-
|
| 2 |
+
import os
|
| 3 |
+
import logging
|
| 4 |
+
import sys
|
| 5 |
+
import gradio as gr
|
| 6 |
+
import torch
|
| 7 |
+
from app_modules.utils import *
|
| 8 |
+
from app_modules.presets import *
|
| 9 |
+
from app_modules.overwrites import *
|
| 10 |
+
|
| 11 |
+
logging.basicConfig(
|
| 12 |
+
level=logging.DEBUG,
|
| 13 |
+
format="%(asctime)s [%(levelname)s] [%(filename)s:%(lineno)d] %(message)s",
|
| 14 |
+
)
|
| 15 |
+
|
| 16 |
+
base_model = "decapoda-research/llama-7b-hf"
|
| 17 |
+
adapter_model = "project-baize/baize-lora-7B"
|
| 18 |
+
#tokenizer,model,device = load_tokenizer_and_model(base_model,adapter_model)
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
def predict(text,
|
| 22 |
+
chatbot,
|
| 23 |
+
history,
|
| 24 |
+
top_p,
|
| 25 |
+
temperature,
|
| 26 |
+
max_length_tokens,
|
| 27 |
+
max_context_length_tokens,):
|
| 28 |
+
if text=="":
|
| 29 |
+
return history,history,"Empty Context"
|
| 30 |
+
|
| 31 |
+
inputs = generate_prompt_with_history(text,history,tokenizer,max_length=max_context_length_tokens)
|
| 32 |
+
if inputs is False:
|
| 33 |
+
return [[x[0],convert_to_markdown(x[1])] for x in history]+[[text,"Sorry, the input is too long."]],history,"Generate Fail"
|
| 34 |
+
else:
|
| 35 |
+
prompt,inputs=inputs
|
| 36 |
+
begin_length = len(prompt)
|
| 37 |
+
|
| 38 |
+
input_ids = inputs["input_ids"].to(device)
|
| 39 |
+
|
| 40 |
+
with torch.no_grad():
|
| 41 |
+
for x in greedy_search(input_ids,model,tokenizer,stop_words=["[|Human|]", "[|AI|]"],max_length=max_length_tokens,temperature=temperature,top_p=top_p):
|
| 42 |
+
if is_stop_word_or_prefix(x,["[|Human|]", "[|AI|]"]) is False:
|
| 43 |
+
if "[|Human|]" in x:
|
| 44 |
+
x = x[:x.index("[|Human|]")].strip()
|
| 45 |
+
if "[|AI|]" in x:
|
| 46 |
+
x = x[:x.index("[|AI|]")].strip()
|
| 47 |
+
x = x.strip(" ")
|
| 48 |
+
a, b= [[y[0],convert_to_markdown(y[1])] for y in history]+[[text, convert_to_markdown(x)]],history + [[text,x]]
|
| 49 |
+
yield a, b, "Generating……"
|
| 50 |
+
if shared_state.interrupted:
|
| 51 |
+
shared_state.recover()
|
| 52 |
+
try:
|
| 53 |
+
yield a, b, "Stop Success"
|
| 54 |
+
return
|
| 55 |
+
except:
|
| 56 |
+
pass
|
| 57 |
+
print(prompt)
|
| 58 |
+
print(x)
|
| 59 |
+
print("="*80)
|
| 60 |
+
try:
|
| 61 |
+
yield a,b,"Generate Success"
|
| 62 |
+
except:
|
| 63 |
+
pass
|
| 64 |
+
|
| 65 |
+
def retry(
|
| 66 |
+
text,
|
| 67 |
+
chatbot,
|
| 68 |
+
history,
|
| 69 |
+
top_p,
|
| 70 |
+
temperature,
|
| 71 |
+
max_length_tokens,
|
| 72 |
+
max_context_length_tokens,
|
| 73 |
+
):
|
| 74 |
+
logging.info("Retry……")
|
| 75 |
+
if len(history) == 0:
|
| 76 |
+
yield chatbot, history, f"Empty context"
|
| 77 |
+
return
|
| 78 |
+
chatbot.pop()
|
| 79 |
+
inputs = history.pop()[0]
|
| 80 |
+
for x in predict(inputs,chatbot,history,top_p,temperature,max_length_tokens,max_context_length_tokens):
|
| 81 |
+
yield x
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
gr.Chatbot.postprocess = postprocess
|
| 85 |
+
|
| 86 |
+
with open("assets/custom.css", "r", encoding="utf-8") as f:
|
| 87 |
+
customCSS = f.read()
|
| 88 |
+
|
| 89 |
+
with gr.Blocks(css=customCSS, theme=small_and_beautiful_theme) as demo:
|
| 90 |
+
history = gr.State([])
|
| 91 |
+
user_question = gr.State("")
|
| 92 |
+
with gr.Row():
|
| 93 |
+
gr.HTML(title)
|
| 94 |
+
status_display = gr.Markdown("Success", elem_id="status_display")
|
| 95 |
+
gr.Markdown(description_top)
|
| 96 |
+
with gr.Row(scale=1).style(equal_height=True):
|
| 97 |
+
with gr.Column(scale=5):
|
| 98 |
+
with gr.Row(scale=1):
|
| 99 |
+
chatbot = gr.Chatbot(elem_id="chuanhu_chatbot").style(height="80%")
|
| 100 |
+
with gr.Row(scale=1):
|
| 101 |
+
with gr.Column(scale=12):
|
| 102 |
+
user_input = gr.Textbox(
|
| 103 |
+
show_label=False, placeholder="Enter text"
|
| 104 |
+
).style(container=False)
|
| 105 |
+
with gr.Column(min_width=70, scale=1):
|
| 106 |
+
submitBtn = gr.Button("Send")
|
| 107 |
+
with gr.Column(min_width=70, scale=1):
|
| 108 |
+
cancelBtn = gr.Button("Stop")
|
| 109 |
+
|
| 110 |
+
with gr.Row(scale=1):
|
| 111 |
+
emptyBtn = gr.Button(
|
| 112 |
+
"🧹 New Conversation",
|
| 113 |
+
)
|
| 114 |
+
retryBtn = gr.Button("🔄 Regenerate")
|
| 115 |
+
delLastBtn = gr.Button("🗑️ Remove Last Turn")
|
| 116 |
+
with gr.Column():
|
| 117 |
+
with gr.Column(min_width=50, scale=1):
|
| 118 |
+
with gr.Tab(label="Parameter Setting"):
|
| 119 |
+
gr.Markdown("# Parameters")
|
| 120 |
+
top_p = gr.Slider(
|
| 121 |
+
minimum=-0,
|
| 122 |
+
maximum=1.0,
|
| 123 |
+
value=0.95,
|
| 124 |
+
step=0.05,
|
| 125 |
+
interactive=True,
|
| 126 |
+
label="Top-p",
|
| 127 |
+
)
|
| 128 |
+
temperature = gr.Slider(
|
| 129 |
+
minimum=-0,
|
| 130 |
+
maximum=2.0,
|
| 131 |
+
value=1,
|
| 132 |
+
step=0.1,
|
| 133 |
+
interactive=True,
|
| 134 |
+
label="Temperature",
|
| 135 |
+
)
|
| 136 |
+
max_length_tokens = gr.Slider(
|
| 137 |
+
minimum=0,
|
| 138 |
+
maximum=512,
|
| 139 |
+
value=512,
|
| 140 |
+
step=8,
|
| 141 |
+
interactive=True,
|
| 142 |
+
label="Max Generation Tokens",
|
| 143 |
+
)
|
| 144 |
+
max_context_length_tokens = gr.Slider(
|
| 145 |
+
minimum=0,
|
| 146 |
+
maximum=4096,
|
| 147 |
+
value=2048,
|
| 148 |
+
step=128,
|
| 149 |
+
interactive=True,
|
| 150 |
+
label="Max History Tokens",
|
| 151 |
+
)
|
| 152 |
+
gr.Markdown(description)
|
| 153 |
+
|
| 154 |
+
predict_args = dict(
|
| 155 |
+
fn=predict,
|
| 156 |
+
inputs=[
|
| 157 |
+
user_question,
|
| 158 |
+
chatbot,
|
| 159 |
+
history,
|
| 160 |
+
top_p,
|
| 161 |
+
temperature,
|
| 162 |
+
max_length_tokens,
|
| 163 |
+
max_context_length_tokens,
|
| 164 |
+
],
|
| 165 |
+
outputs=[chatbot, history, status_display],
|
| 166 |
+
show_progress=True,
|
| 167 |
+
)
|
| 168 |
+
retry_args = dict(
|
| 169 |
+
fn=retry,
|
| 170 |
+
inputs=[
|
| 171 |
+
user_input,
|
| 172 |
+
chatbot,
|
| 173 |
+
history,
|
| 174 |
+
top_p,
|
| 175 |
+
temperature,
|
| 176 |
+
max_length_tokens,
|
| 177 |
+
max_context_length_tokens,
|
| 178 |
+
],
|
| 179 |
+
outputs=[chatbot, history, status_display],
|
| 180 |
+
show_progress=True,
|
| 181 |
+
)
|
| 182 |
+
|
| 183 |
+
reset_args = dict(
|
| 184 |
+
fn=reset_textbox, inputs=[], outputs=[user_input, status_display]
|
| 185 |
+
)
|
| 186 |
+
|
| 187 |
+
# Chatbot
|
| 188 |
+
cancelBtn.click(cancel_outputing, [], [ status_display])
|
| 189 |
+
transfer_input_args = dict(
|
| 190 |
+
fn=transfer_input, inputs=[user_input], outputs=[user_question, user_input, submitBtn, cancelBtn], show_progress=True
|
| 191 |
+
)
|
| 192 |
+
|
| 193 |
+
user_input.submit(**transfer_input_args).then(**predict_args)
|
| 194 |
+
|
| 195 |
+
submitBtn.click(**transfer_input_args).then(**predict_args)
|
| 196 |
+
|
| 197 |
+
emptyBtn.click(
|
| 198 |
+
reset_state,
|
| 199 |
+
outputs=[chatbot, history, status_display],
|
| 200 |
+
show_progress=True,
|
| 201 |
+
)
|
| 202 |
+
emptyBtn.click(**reset_args)
|
| 203 |
+
|
| 204 |
+
retryBtn.click(**retry_args)
|
| 205 |
+
|
| 206 |
+
delLastBtn.click(
|
| 207 |
+
delete_last_conversation,
|
| 208 |
+
[chatbot, history],
|
| 209 |
+
[chatbot, history, status_display],
|
| 210 |
+
show_progress=True,
|
| 211 |
+
)
|
| 212 |
+
|
| 213 |
+
demo.title = "Baize"
|
| 214 |
+
|
| 215 |
+
if __name__ == "__main__":
|
| 216 |
+
reload_javascript()
|
| 217 |
+
demo.queue(concurrency_count=CONCURRENT_COUNT).launch(
|
| 218 |
+
share=True, favicon_path="./assets/favicon.ico", inbrowser=True
|
| 219 |
+
)
|
requirements.txt
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio
|
| 2 |
+
mdtex2html
|
| 3 |
+
pypinyin
|
| 4 |
+
tiktoken
|
| 5 |
+
socksio
|
| 6 |
+
tqdm
|
| 7 |
+
colorama
|
| 8 |
+
duckduckgo_search
|
| 9 |
+
Pygments
|
| 10 |
+
llama_index
|
| 11 |
+
langchain
|
| 12 |
+
markdown
|
| 13 |
+
markdown2
|
| 14 |
+
|