data_gen / app.py
ramalMr's picture
Update app.py
3c1274a verified
raw
history blame
3.17 kB
from huggingface_hub import InferenceClient
import gradio as gr
import PyPDF2
import random
import pandas as pd
from io import StringIO
# Initialize the inference client with your chosen model
client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
def extract_text_from_pdf(file):
pdf_reader = PyPDF2.PdfReader(file)
text = ""
for page in range(len(pdf_reader.pages)):
text += pdf_reader.pages[page].extract_text()
return text
def generate_synthetic_data(sentences, temperature, max_new_tokens, top_p, repetition_penalty):
synthetic_data = []
for sentence in sentences:
# Trim whitespace and skip if the sentence is empty
sentence = sentence.strip()
if not sentence:
continue
generate_kwargs = {
"temperature": temperature,
"max_new_tokens": max_new_tokens,
"top_p": top_p,
"repetition_penalty": repetition_penalty,
"do_sample": True,
"seed": 42,
}
try:
stream = client.text_generation(sentence, **generate_kwargs, stream=True, details=True, return_full_text=False)
output = ""
for response in stream:
output += response.token.text
synthetic_data.append(output)
except Exception as e:
print(f"Error generating data for sentence '{sentence}': {e}")
# Optionally, append a placeholder or error message to `synthetic_data` to maintain alignment with input sentences
synthetic_data.append(f"Error: {e}")
return synthetic_data
def generate(file, temperature, max_new_tokens, top_p, repetition_penalty):
# Extract text and split into sentences
text = extract_text_from_pdf(file)
sentences = text.split('.')
random.shuffle(sentences) # Shuffle sentences
synthetic_data = generate_synthetic_data(sentences, temperature, max_new_tokens, top_p, repetition_penalty)
# Convert synthetic data to CSV
df = pd.DataFrame(synthetic_data, columns=["Synthetic Data"])
csv_buffer = StringIO()
df.to_csv(csv_buffer, index=False)
return gr.File(value=csv_buffer.getvalue(), file_name="synthetic_data.csv")
gr.Interface(
fn=generate,
inputs=[
gr.File(label="Upload PDF File", file_count="single", file_types=[".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=256, minimum=0, maximum=5120, step=64, interactive=True, info="The maximum numbers of new tokens"),
gr.Slider(label="Top-p (nucleus sampling)", value=0.95, minimum=0.0, maximum=1, step=0.05, interactive=True, info="Higher values sample more low-probability tokens"),
gr.Slider(label="Repetition penalty", value=1.0, minimum=1.0, maximum=2.0, step=0.1, interactive=True, info="Penalize repeated tokens"),
],
outputs="file",
title="Synthetic Data Generation",
description="This tool generates synthetic data from the sentences in your PDF.",
allow_flagging="never",
).launch()