Spaces:
Running
on
Zero
Running
on
Zero
File size: 13,902 Bytes
2317674 1cfe513 2317674 0cdbe8f 2317674 f4a0f87 1cfe513 f4a0f87 a188372 f4a0f87 1cfe513 f4a0f87 1cfe513 0b984ea 1cfe513 f4a0f87 1cfe513 f4a0f87 2317674 1cfe513 2317674 1cfe513 2317674 1cfe513 2317674 1cfe513 2317674 1cfe513 2317674 1cfe513 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 |
import os
from dotenv import load_dotenv
import gradio as gr
from huggingface_hub import InferenceClient
import pandas as pd
import json
from datetime import datetime
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
# ํ๊ฒฝ ๋ณ์ ์ค์
HF_TOKEN = os.getenv("HF_TOKEN")
MODEL_ID = "CohereForAI/c4ai-command-r7b-12-2024"
from transformers import pipeline
class ModelManager:
def __init__(self):
self.pipe = None
self.setup_pipeline()
def setup_pipeline(self):
try:
print("ํ์ดํ๋ผ์ธ ์ด๊ธฐํ ์์...")
self.pipe = pipeline(
"text-generation",
model=MODEL_ID,
token=HF_TOKEN,
device_map="auto",
torch_dtype=torch.float16
)
print("ํ์ดํ๋ผ์ธ ์ด๊ธฐํ ์๋ฃ")
except Exception as e:
print(f"ํ์ดํ๋ผ์ธ ์ด๊ธฐํ ์ค ์ค๋ฅ ๋ฐ์: {e}")
raise Exception(f"ํ์ดํ๋ผ์ธ ์ด๊ธฐํ ์คํจ: {e}")
def generate_response(self, messages, max_tokens=4000, temperature=0.7, top_p=0.9):
try:
# ๋ฉ์์ง ํ์ ๋ณํ
prompt = ""
for msg in messages:
role = msg["role"]
content = msg["content"]
if role == "system":
prompt += f"System: {content}\n"
elif role == "user":
prompt += f"User: {content}\n"
elif role == "assistant":
prompt += f"Assistant: {content}\n"
# ์๋ต ์์ฑ
response = self.pipe(
prompt,
max_new_tokens=max_tokens,
temperature=temperature,
top_p=top_p,
do_sample=True,
num_return_sequences=1,
pad_token_id=self.pipe.tokenizer.eos_token_id
)
# ์๋ต ํ
์คํธ ์ถ์ถ ๋ฐ ์คํธ๋ฆฌ๋ฐ ์๋ฎฌ๋ ์ด์
generated_text = response[0]['generated_text'][len(prompt):].strip()
words = generated_text.split()
# ๋จ์ด ๋จ์๋ก ์คํธ๋ฆฌ๋ฐ
partial_response = ""
for word in words:
partial_response += word + " "
yield type('Response', (), {
'choices': [type('Choice', (), {
'delta': {'content': word + " "}
})()]
})()
except Exception as e:
raise Exception(f"์๋ต ์์ฑ ์คํจ: {e}")
class ChatHistory:
def __init__(self):
self.history = []
self.history_file = "/tmp/chat_history.json"
self.load_history()
def add_conversation(self, user_msg: str, assistant_msg: str):
conversation = {
"timestamp": datetime.now().isoformat(),
"messages": [
{"role": "user", "content": user_msg},
{"role": "assistant", "content": assistant_msg}
]
}
self.history.append(conversation)
self.save_history()
def format_for_display(self):
formatted = []
for conv in self.history:
formatted.append([
conv["messages"][0]["content"],
conv["messages"][1]["content"]
])
return formatted
def get_messages_for_api(self):
messages = []
for conv in self.history:
messages.extend([
{"role": "user", "content": conv["messages"][0]["content"]},
{"role": "assistant", "content": conv["messages"][1]["content"]}
])
return messages
def clear_history(self):
self.history = []
self.save_history()
def save_history(self):
try:
with open(self.history_file, 'w', encoding='utf-8') as f:
json.dump(self.history, f, ensure_ascii=False, indent=2)
except Exception as e:
print(f"ํ์คํ ๋ฆฌ ์ ์ฅ ์คํจ: {e}")
def load_history(self):
try:
if os.path.exists(self.history_file):
with open(self.history_file, 'r', encoding='utf-8') as f:
self.history = json.load(f)
except Exception as e:
print(f"ํ์คํ ๋ฆฌ ๋ก๋ ์คํจ: {e}")
self.history = []
# ์ ์ญ ์ธ์คํด์ค ์์ฑ
chat_history = ChatHistory()
model_manager = ModelManager()
def get_client():
return InferenceClient(MODEL_ID, token=HF_TOKEN)
def analyze_file_content(content, file_type):
"""Analyze file content and return structural summary"""
if file_type in ['parquet', 'csv']:
try:
lines = content.split('\n')
header = lines[0]
columns = header.count('|') - 1
rows = len(lines) - 3
return f"๐ ๋ฐ์ดํฐ์
๊ตฌ์กฐ: {columns}๊ฐ ์ปฌ๋ผ, {rows}๊ฐ ๋ฐ์ดํฐ"
except:
return "โ ๋ฐ์ดํฐ์
๊ตฌ์กฐ ๋ถ์ ์คํจ"
lines = content.split('\n')
total_lines = len(lines)
non_empty_lines = len([line for line in lines if line.strip()])
if any(keyword in content.lower() for keyword in ['def ', 'class ', 'import ', 'function']):
functions = len([line for line in lines if 'def ' in line])
classes = len([line for line in lines if 'class ' in line])
imports = len([line for line in lines if 'import ' in line or 'from ' in line])
return f"๐ป ์ฝ๋ ๊ตฌ์กฐ: {total_lines}์ค (ํจ์: {functions}, ํด๋์ค: {classes}, ์ํฌํธ: {imports})"
paragraphs = content.count('\n\n') + 1
words = len(content.split())
return f"๐ ๋ฌธ์ ๊ตฌ์กฐ: {total_lines}์ค, {paragraphs}๋จ๋ฝ, ์ฝ {words}๋จ์ด"
def read_uploaded_file(file):
if file is None:
return "", ""
try:
file_ext = os.path.splitext(file.name)[1].lower()
if file_ext == '.parquet':
df = pd.read_parquet(file.name, engine='pyarrow')
content = df.head(10).to_markdown(index=False)
return content, "parquet"
elif file_ext == '.csv':
encodings = ['utf-8', 'cp949', 'euc-kr', 'latin1']
for encoding in encodings:
try:
df = pd.read_csv(file.name, encoding=encoding)
content = f"๐ ๋ฐ์ดํฐ ๋ฏธ๋ฆฌ๋ณด๊ธฐ:\n{df.head(10).to_markdown(index=False)}\n\n"
content += f"\n๐ ๋ฐ์ดํฐ ์ ๋ณด:\n"
content += f"- ์ ์ฒด ํ ์: {len(df)}\n"
content += f"- ์ ์ฒด ์ด ์: {len(df.columns)}\n"
content += f"- ์ปฌ๋ผ ๋ชฉ๋ก: {', '.join(df.columns)}\n"
content += f"\n๐ ์ปฌ๋ผ ๋ฐ์ดํฐ ํ์
:\n"
for col, dtype in df.dtypes.items():
content += f"- {col}: {dtype}\n"
null_counts = df.isnull().sum()
if null_counts.any():
content += f"\nโ ๏ธ ๊ฒฐ์ธก์น:\n"
for col, null_count in null_counts[null_counts > 0].items():
content += f"- {col}: {null_count}๊ฐ ๋๋ฝ\n"
return content, "csv"
except UnicodeDecodeError:
continue
raise UnicodeDecodeError(f"โ ์ง์๋๋ ์ธ์ฝ๋ฉ์ผ๋ก ํ์ผ์ ์ฝ์ ์ ์์ต๋๋ค ({', '.join(encodings)})")
else:
encodings = ['utf-8', 'cp949', 'euc-kr', 'latin1']
for encoding in encodings:
try:
with open(file.name, 'r', encoding=encoding) as f:
content = f.read()
return content, "text"
except UnicodeDecodeError:
continue
raise UnicodeDecodeError(f"โ ์ง์๋๋ ์ธ์ฝ๋ฉ์ผ๋ก ํ์ผ์ ์ฝ์ ์ ์์ต๋๋ค ({', '.join(encodings)})")
except Exception as e:
return f"โ ํ์ผ ์ฝ๊ธฐ ์ค๋ฅ: {str(e)}", "error"
def chat(message, history, uploaded_file, system_message="", max_tokens=4000, temperature=0.7, top_p=0.9):
if not message:
return "", history
system_prefix = """์ ๋ ์ฌ๋ฌ๋ถ์ ์น๊ทผํ๊ณ ์ง์ ์ธ AI ์ด์์คํดํธ 'GiniGEN'์
๋๋ค.. ๋ค์๊ณผ ๊ฐ์ ์์น์ผ๋ก ์ํตํ๊ฒ ์ต๋๋ค:
1. ๐ค ์น๊ทผํ๊ณ ๊ณต๊ฐ์ ์ธ ํ๋๋ก ๋ํ
2. ๐ก ๋ช
ํํ๊ณ ์ดํดํ๊ธฐ ์ฌ์ด ์ค๋ช
์ ๊ณต
3. ๐ฏ ์ง๋ฌธ์ ์๋๋ฅผ ์ ํํ ํ์
ํ์ฌ ๋ง์ถคํ ๋ต๋ณ
4. ๐ ํ์ํ ๊ฒฝ์ฐ ์
๋ก๋๋ ํ์ผ ๋ด์ฉ์ ์ฐธ๊ณ ํ์ฌ ๊ตฌ์ฒด์ ์ธ ๋์ ์ ๊ณต
5. โจ ์ถ๊ฐ์ ์ธ ํต์ฐฐ๊ณผ ์ ์์ ํตํ ๊ฐ์น ์๋ ๋ํ
ํญ์ ์์ ๋ฐ๋ฅด๊ณ ์น์ ํ๊ฒ ์๋ตํ๋ฉฐ, ํ์ํ ๊ฒฝ์ฐ ๊ตฌ์ฒด์ ์ธ ์์๋ ์ค๋ช
์ ์ถ๊ฐํ์ฌ
์ดํด๋ฅผ ๋๊ฒ ์ต๋๋ค."""
try:
if uploaded_file:
content, file_type = read_uploaded_file(uploaded_file)
if file_type == "error":
error_message = content
chat_history.add_conversation(message, error_message)
return "", history + [[message, error_message]]
file_summary = analyze_file_content(content, file_type)
if file_type in ['parquet', 'csv']:
system_message += f"\n\nํ์ผ ๋ด์ฉ:\n```markdown\n{content}\n```"
else:
system_message += f"\n\nํ์ผ ๋ด์ฉ:\n```\n{content}\n```"
if message == "ํ์ผ ๋ถ์์ ์์ํฉ๋๋ค...":
message = f"""[ํ์ผ ๊ตฌ์กฐ ๋ถ์] {file_summary}
๋ค์ ๊ด์ ์์ ๋์์ ๋๋ฆฌ๊ฒ ์ต๋๋ค:
1. ๐ ์ ๋ฐ์ ์ธ ๋ด์ฉ ํ์
2. ๐ก ์ฃผ์ ํน์ง ์ค๋ช
3. ๐ฏ ์ค์ฉ์ ์ธ ํ์ฉ ๋ฐฉ์
4. โจ ๊ฐ์ ์ ์
5. ๐ฌ ์ถ๊ฐ ์ง๋ฌธ์ด๋ ํ์ํ ์ค๋ช
"""
messages = [{"role": "system", "content": system_prefix + system_message}]
if history:
for user_msg, assistant_msg in history:
messages.append({"role": "user", "content": user_msg})
messages.append({"role": "assistant", "content": assistant_msg})
messages.append({"role": "user", "content": message})
client = get_client()
partial_message = ""
for msg in client.chat_completion(
messages,
max_tokens=max_tokens,
stream=True,
temperature=temperature,
top_p=top_p,
):
token = msg.choices[0].delta.get('content', None)
if token:
partial_message += token
current_history = history + [[message, partial_message]]
yield "", current_history
chat_history.add_conversation(message, partial_message)
except Exception as e:
error_msg = f"โ ์ค๋ฅ๊ฐ ๋ฐ์ํ์ต๋๋ค: {str(e)}"
chat_history.add_conversation(message, error_msg)
yield "", history + [[message, error_msg]]
with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", title="GiniGEN ๐ค") as demo:
initial_history = chat_history.format_for_display()
with gr.Row():
with gr.Column(scale=2):
chatbot = gr.Chatbot(
value=initial_history,
height=600,
label="๋ํ์ฐฝ ๐ฌ",
show_label=True
)
msg = gr.Textbox(
label="๋ฉ์์ง ์
๋ ฅ",
show_label=False,
placeholder="๋ฌด์์ด๋ ๋ฌผ์ด๋ณด์ธ์... ๐ญ",
container=False
)
with gr.Row():
clear = gr.ClearButton([msg, chatbot], value="๋ํ๋ด์ฉ ์ง์ฐ๊ธฐ")
send = gr.Button("๋ณด๋ด๊ธฐ ๐ค")
with gr.Column(scale=1):
gr.Markdown("### GiniGEN ๐ค [ํ์ผ ์
๋ก๋] ๐\n์ง์ ํ์: ํ
์คํธ, ์ฝ๋, CSV, Parquet ํ์ผ")
file_upload = gr.File(
label="ํ์ผ ์ ํ",
file_types=["text", ".csv", ".parquet"],
type="filepath"
)
with gr.Accordion("๊ณ ๊ธ ์ค์ โ๏ธ", open=False):
system_message = gr.Textbox(label="์์คํ
๋ฉ์์ง ๐", value="")
max_tokens = gr.Slider(minimum=1, maximum=8000, value=4000, label="์ต๋ ํ ํฐ ์ ๐")
temperature = gr.Slider(minimum=0, maximum=1, value=0.7, label="์ฐฝ์์ฑ ์์ค ๐ก๏ธ")
top_p = gr.Slider(minimum=0, maximum=1, value=0.9, label="์๋ต ๋ค์์ฑ ๐")
gr.Examples(
examples=[
["์๋
ํ์ธ์! ์ด๋ค ๋์์ด ํ์ํ์ ๊ฐ์? ๐ค"],
["์ ๊ฐ ์ดํดํ๊ธฐ ์ฝ๊ฒ ์ค๋ช
ํด ์ฃผ์๊ฒ ์ด์? ๐"],
["์ด ๋ด์ฉ์ ์ค์ ๋ก ์ด๋ป๊ฒ ํ์ฉํ ์ ์์๊น์? ๐ฏ"],
["์ถ๊ฐ๋ก ์กฐ์ธํด ์ฃผ์ค ๋ด์ฉ์ด ์์ผ์ ๊ฐ์? โจ"],
["๊ถ๊ธํ ์ ์ด ๋ ์๋๋ฐ ์ฌ์ญค๋ด๋ ๋ ๊น์? ๐ค"],
],
inputs=msg,
)
def clear_chat():
chat_history.clear_history()
return None, None
msg.submit(
chat,
inputs=[msg, chatbot, file_upload, system_message, max_tokens, temperature, top_p],
outputs=[msg, chatbot]
)
send.click(
chat,
inputs=[msg, chatbot, file_upload, system_message, max_tokens, temperature, top_p],
outputs=[msg, chatbot]
)
clear.click(
clear_chat,
outputs=[msg, chatbot]
)
file_upload.change(
lambda: "ํ์ผ ๋ถ์์ ์์ํฉ๋๋ค...",
outputs=msg
).then(
chat,
inputs=[msg, chatbot, file_upload, system_message, max_tokens, temperature, top_p],
outputs=[msg, chatbot]
)
if __name__ == "__main__":
demo.launch() |