File size: 12,988 Bytes
826babd e0a3af3 826babd 4d24f6b 826babd 6daa349 826babd 4d24f6b 826babd e0a3af3 826babd |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 |
#from haystack.components.generators import HuggingFaceTGIGenerator
from llama_index.llms import HuggingFaceInferenceAPI
from llama_index.llms import ChatMessage, MessageRole
from llama_index.prompts import ChatPromptTemplate
from llama_index import VectorStoreIndex, SimpleDirectoryReader, ServiceContext #, LLMPredictor, StorageContext, load_index_from_storage
import gradio as gr
#import sys
#import logging
#import torch
#from huggingface_hub import InferenceClient
#import tqdm as notebook_tqdm
import requests
import os
import json
#generator = HuggingFaceTGIGenerator("mistralai/Mixtral-8x7B-Instruct-v0.1")
#generator.warm_up()
def download_file(url, filename):
"""
Download a file from the specified URL and save it locally under the given filename.
"""
response = requests.get(url, stream=True)
# Check if the request was successful
if filename in os.listdir('content/'): return
if filename == '': return
if response.status_code == 200:
with open('content/' + filename, 'wb') as file:
for chunk in response.iter_content(chunk_size=1024):
if chunk: # filter out keep-alive new chunks
file.write(chunk)
print(f"Download complete: {filename}")
else:
print(f"Error: Unable to download file. HTTP status code: {response.status_code}")
#def save_answer(prompt, rag_answer, norag_answer):
# json_dict = dict()
# json_dict['prompt'] = prompt
# json_dict['rag_answer'] = rag_answer
# json_dict['norag_answer'] = norag_answer
#
# file_path = 'saved_answers.json'
#
# # Check if the file exists
# if not os.path.isfile(file_path):
# with open(file_path, 'w') as f:
# # Create an empty list in the file to store dictionaries
# json.dump([], f)
# f.write('\n') # Add a newline to separate the list and future entries
#
# # Open the file in append mode
# with open(file_path, 'a+') as f:
# # Read the existing data
# f.seek(0)
# data = json.load(f)
#
# # Append the new dictionary to the list
# data.append(json_dict)
#
# # Move the cursor to the beginning of the file
# f.seek(0)
#
# # Write the updated list of dictionaries
# json.dump(data, f)
# f.write('\n') # Add a newline to separate the list and future entries
#
#
#def check_answer(prompt):
# file_path = 'saved_answers.json'
#
# if not os.path.isfile(file_path):
# with open(file_path, 'w') as f:
# # Create an empty list in the file to store dictionaries
# json.dump([], f)
# f.write('\n') # Add a newline to separate the list and future entries
# with open('saved_answers.json', 'r') as f:
# data = json.load(f)
# for entry in data:
# if entry['prompt'] == prompt:
# return entry['rag_answer'], entry['norag_answer']
# return None, None # Return None if the prompt is not found
def save_answer(prompt, rag_answer, norag_answer):
file_path = 'saved_answers.jsonl'
# Create a dictionary for the current answer
json_dict = {
'prompt': prompt,
'rag_answer': rag_answer,
'norag_answer': norag_answer
}
# Check if the file exists, and create it if not
#if not os.path.isfile(file_path):
# with open(file_path, 'w') as f:
# # Create an empty list in the file to store dictionaries
# json.dump([], f)
# f.write('\n') # Add a newline to separate the list and future entries
# Load existing data from the file
existing_data = load_jsonl(file_path)
# Append the new answer to the existing data
existing_data.append(json_dict)
# Save the updated data back to the file
write_to_jsonl(file_path, existing_data)
def check_answer(prompt):
file_path = 'saved_answers.jsonl'
## Check if the file exists, and create it if not
#if not os.path.isfile(file_path):
# with open(file_path, 'w') as f:
# # Create an empty list in the file to store dictionaries
# json.dump([], f)
# f.write('\n') # Add a newline to separate the list and future entries
# Load existing data from the file
try:
existing_data = load_jsonl(file_path)
except:
return None, None
if len(existing_data) == 0:
return None, None
# Find the answer for the given prompt, if it exists
for entry in existing_data:
if entry['prompt'] == prompt:
return entry['rag_answer'], entry['norag_answer']
# Return None if the prompt is not found
return None, None
# Helper functions
def load_jsonl(file_path):
data = []
with open(file_path, 'r') as file:
for line in file:
# Each line is a JSON object
item = json.loads(line)
data.append(item)
return data
def write_to_jsonl(file_path, data):
with open(file_path, 'a+') as file:
for item in data:
# Convert Python object to JSON string and write it to the file
json_line = json.dumps(item)
file.write(json_line + '\n')
def generate(prompt, history, rag_only, file_link, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,):
rag_answer, norag_answer = check_answer(prompt)
if rag_answer != None:
if rag_only:
return f'* Mixtral + RAG Output:\n{rag_answer}'
else:
return f'* Mixtral Output:\n{norag_answer}\n\n* Mixtral + RAG Output:\n{rag_answer}'
mixtral = HuggingFaceInferenceAPI(
model_name="mistralai/Mixtral-8x7B-Instruct-v0.1"
#Mistral-7B-Instruct-v0.2
)
service_context = ServiceContext.from_defaults(
llm=mixtral, embed_model="local:BAAI/bge-small-en-v1.5"
)
download = download_file(file_link,file_link.split("/")[-1])
documents = SimpleDirectoryReader("content/").load_data()
index = VectorStoreIndex.from_documents(documents,service_context=service_context)
# Text QA Prompt
chat_text_qa_msgs = [
ChatMessage(
role=MessageRole.SYSTEM,
content=(
"Always answer the question, even if the context isn't helpful."
),
),
ChatMessage(
role=MessageRole.USER,
content=(
"Context information is below.\n"
"---------------------\n"
"{context_str}\n"
"---------------------\n"
"Given the context information and not prior knowledge, "
"answer the question: {query_str}\n"
),
),
]
text_qa_template = ChatPromptTemplate(chat_text_qa_msgs)
# Refine Prompt
chat_refine_msgs = [
ChatMessage(
role=MessageRole.SYSTEM,
content=(
"Always answer the question, even if the context isn't helpful."
),
),
ChatMessage(
role=MessageRole.USER,
content=(
"We have the opportunity to refine the original answer "
"(only if needed) with some more context below.\n"
"------------\n"
"{context_msg}\n"
"------------\n"
"Given the new context, refine the original answer to better "
"answer the question: {query_str}. "
"If the context isn't useful, output the original answer again.\n"
"Original Answer: {existing_answer}"
),
),
]
refine_template = ChatPromptTemplate(chat_refine_msgs)
temperature = float(temperature)
if temperature < 1e-2:
temperature = 1e-2
top_p = float(top_p)
stream= index.as_query_engine(
text_qa_template=text_qa_template, refine_template=refine_template, similarity_top_k=6, temperature = temperature,
max_new_tokens=max_new_tokens, top_p=top_p, repetition_penalty = repetition_penalty
).query(prompt)
print(str(stream))
output_rag= str(stream) #""
#output_norag = mixtral.complete(prompt, details=True, similarity_top_k=6, temperature = temperature,
# max_new_tokens=max_new_tokens, top_p=top_p, repetition_penalty = repetition_penalty)
#for response in str(stream):
# output += response
# yield output
#print(output_norag)
#result = generator.run(prompt, generation_kwargs={"max_new_tokens": 350})
#output_norag = result["replies"][0]
### NORAG
if rag_only == False:
chat_text_qa_msgs_nr = [
ChatMessage(
role=MessageRole.SYSTEM,
content=(
"Always answer the question"
),
),
ChatMessage(
role=MessageRole.USER,
content=(
"answer the question: {query_str}\n"
),
),
]
text_qa_template_nr = ChatPromptTemplate(chat_text_qa_msgs_nr)
# Refine Prompt
chat_refine_msgs_nr = [
ChatMessage(
role=MessageRole.SYSTEM,
content=(
"Always answer the question"
),
),
ChatMessage(
role=MessageRole.USER,
content=(
"answer the question: {query_str}. "
"If the context isn't useful, output the original answer again.\n"
"Original Answer: {existing_answer}"
),
),
]
refine_template_nr = ChatPromptTemplate(chat_refine_msgs_nr)
stream_nr= index.as_query_engine(
text_qa_template=text_qa_template_nr, refine_template=refine_template_nr, similarity_top_k=6
).query(prompt)
###
output_norag = str(stream_nr)
save_answer(prompt, output_rag, output_norag)
return f'* Mixtral Output:\n{output_norag}\n\n* Mixtral + RAG Output:\n{output_rag}'
return f'* Mixtral + RAG Output:\n{output_rag}'
#for response in formatted_output:
# output += response
# yield output
#return formatted_output
def upload_file(files):
file_paths = [file.name for file in files]
return file_paths
additional_inputs=[
gr.Checkbox(
label="RAG Only",
interactive=True,
value= False
),
gr.Textbox(
label="File Link",
max_lines=1,
interactive=True,
value= "https://arxiv.org/pdf/2401.10020.pdf"
),
gr.Slider(
label="Temperature",
value=0.9,
minimum=0.0,
maximum=1.0,
step=0.05,
interactive=True,
info="Higher values produce more diverse outputs",
),
gr.Slider(
label="Max new tokens",
value=1024,
minimum=0,
maximum=2048,
step=64,
interactive=True,
info="The maximum numbers of new tokens",
),
gr.Slider(
label="Top-p (nucleus sampling)",
value=0.90,
minimum=0.0,
maximum=1,
step=0.05,
interactive=True,
info="Higher values sample more low-probability tokens",
),
gr.Slider(
label="Repetition penalty",
value=1.2,
minimum=1.0,
maximum=2.0,
step=0.05,
interactive=True,
info="Penalize repeated tokens",
)
]
examples=[["What is a trustworthy digital repository, where can you find this information?", None, None, None, None, None, None, ],
["What are things a repository must have?", None, None, None, None, None, None,],
["What principles should record creators follow?", None, None, None, None, None, None,],
["Write a very short summary of Data Sanitation Techniques by Edgar Dale, and write a citation in APA style.", None, None, None, None, None, None,],
["Can you explain how the QuickSort algorithm works and provide a Python implementation?", None, None, None, None, None, None,],
["What are some unique features of Rust that make it stand out compared to other systems programming languages like C++?", None, None, None, None, None, None,],
]
gr.ChatInterface(
fn=generate,
chatbot=gr.Chatbot(show_label=False, show_share_button=False, show_copy_button=True, likeable=True, layout="panel"),
additional_inputs=additional_inputs,
title="RAG Demo",
examples=examples,
#concurrency_limit=20,
).queue().launch(show_api=False,debug=True,share=True)
#iface = gr.Interface(fn=generate, inputs=["text"], outputs=["text", "text"],
# additional_inputs=additional_inputs, title="RAG Demo", examples=examples)
#iface.launch()
|