File size: 1,922 Bytes
5ec8c71 ee65db2 5ec8c71 ee65db2 5ec8c71 ee65db2 5ec8c71 ee65db2 5ec8c71 f308447 5ec8c71 |
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 |
import os
import gradio as gr
import torch
from peft import PeftModel, PeftConfig
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
def load_data(file_obj):
"""
Load data from the file object of the gr.File() inputs
"""
path = file_obj.name
with open(path, "r") as f:
data = f.read()
return data
def preprocessing(data):
texts = list()
i = 0
if len(data) <= i+4000:
texts = data
else:
while len(data[i:]) != 0:
if len(data[i:]) > 4000:
string = str(data[i:i+4000])
texts.append(string)
i = i + 3800
else:
string = str(data[i:])
texts.append(string)
break
return texts
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
peft_model_id = "sooolee/flan-t5-base-cnn-samsum-lora"
config = PeftConfig.from_pretrained(peft_model_id)
tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
model = AutoModelForSeq2SeqLM.from_pretrained(config.base_model_name_or_path, device_map='auto') # load_in_8bit=True,
model = PeftModel.from_pretrained(model, peft_model_id, device_map='auto')
def summarize(path):
transcript = load_data(file_obj)
texts = preprocessing(transcript)
inputs = tokenizer(texts, return_tensors="pt", padding=True, )
with torch.no_grad():
output_tokens = model.generate(input_ids=inputs["input_ids"].to("cuda"), max_new_tokens=60, do_sample=True, top_p=0.9)
outputs = tokenizer.batch_decode(output_tokens.detach().cpu().numpy(), skip_special_tokens=True)
return outputs
gr.Interface(
fn=summarize,
title = 'Summarize Transcripts',
inputs = gr.File(file_types=["text"], label="Upload a text file.", interactive=True),
outputs = gr.Textbox(label="Summary", max_lines=120, interactive=False),
).launch()
|