SciAssist / controlled_summarization.py
wing-nus's picture
Update controlled_summarization.py
96bfa0d
raw
history blame
9.31 kB
from typing import List, Tuple
import torch
from SciAssist import Summarization
import os
import requests
from datasets import load_dataset
print(f"Is CUDA available: {torch.cuda.is_available()}")
# True
print(f"CUDA device: {torch.cuda.get_device_name(torch.cuda.current_device())}")
acl_data = load_dataset("dyxohjl666/CocoScisum_ACL", revision="refs/convert/parquet")
device = "gpu" if torch.cuda.is_available() else "cpu"
ctrlsum_pipeline = Summarization(os_name="nt",model_name="flan-t5-xl",checkpoint="dyxohjl666/flant5-xl-cocoscisum",device=device)
acl_dict = {}
recommended_kw = {}
def convert_to_dict(data):
""" Dict:
{ url:
{length:
{keywords: summary};
raw_text:
str;
}
}
"""
url = data["url"]
text = data["text"]
keywords = data["keywords"]
length = data["length"]
summary = data["summary"]
for u, t, k, l, s in zip(url, text, keywords, length, summary):
if len(u) < 5:
continue
u = u + ".pdf"
if k == None:
k = ""
if l == None:
l = ""
k = str(k).strip()
l = str(l).strip()
if u in acl_dict.keys():
if k in acl_dict[u][l].keys():
continue
else:
acl_dict[u][l][k] = s
else:
acl_dict[u] = {"": {}, "50": {}, "100": {}, "200": {}, "raw_text": t}
# kws
if u in recommended_kw.keys():
if k == "" or k in recommended_kw[u]:
continue
else:
recommended_kw[u].append(k)
else:
recommended_kw[u] = []
return 1
for i in acl_data.keys():
signal = convert_to_dict(acl_data[i])
def download_pdf(url, dest_folder):
"""
Download a PDF from a given URL and save it to a specified destination folder.
Parameters:
url (str): URL of the PDF
dest_folder (str): Destination folder to save the downloaded PDF
"""
if not os.path.exists(dest_folder):
os.makedirs(dest_folder)
response = requests.get(url, stream=True)
filename = os.path.join(dest_folder, url.split("/")[-1])
with open(filename, 'wb') as file:
for chunk in response.iter_content(chunk_size=1024):
if chunk:
file.write(chunk)
print(f"Downloaded {url} to {filename}")
return filename
def ctrlsum_for_str(input, length=None, keywords=None) -> List[Tuple[str, str]]:
if keywords is not None:
keywords = keywords.strip().split(",")
if keywords[0] == "":
keywords = None
if length == 0 or length is None:
length = None
results = ctrlsum_pipeline.predict(input, type="str",
length=length, keywords=keywords, num_beams=1)
output = []
for res in results["summary"]:
output.append(f"{res}\n\n")
return "".join(output)
def ctrlsum_for_file(input=None, length=None, keywords="", text="", url="") -> List[Tuple[str, str, str]]:
if input == None and url == "":
if text == "":
return None, "Input cannot be left blank.", None
else:
return ctrlsum_for_str(text, length, keywords), text, None
else:
filename = ""
url = url.strip()
if url != "":
if len(url) > 4 and url[-3:] == "pdf":
if url.strip() in acl_dict.keys():
raw_text = acl_dict[url]["raw_text"]
l = str(length)
if length == 0:
l = ""
if l in acl_dict[url].keys():
if keywords.strip() in acl_dict[url][l].keys():
summary = acl_dict[url][l][keywords]
return summary, raw_text, None
if keywords.strip() == "":
keywords = None
if l == "":
l = None
return ctrlsum_for_str(raw_text, int(l), keywords), raw_text, None
filename = download_pdf(url, './cache/')
else:
"Invalid url(Not PDF)!", None, None
else:
filename = input.name
if keywords != "":
keywords = keywords.strip().split(",")
if keywords[0] == "":
keywords = None
if length == 0:
length = None
# Identify the format of input and parse reference strings
if filename[-4:] == ".txt":
results = ctrlsum_pipeline.predict(filename, type="txt",
save_results=False,
length=length, keywords=keywords, num_beams=1)
elif filename[-4:] == ".pdf":
results = ctrlsum_pipeline.predict(filename,
save_results=False, length=length, keywords=keywords, num_beams=1)
else:
return "File Format Error !", None, filename
output = []
for res in results["summary"]:
output.append(f"{res}\n\n")
return "".join(output), results["raw_text"], filename
ctrlsum_str_example = "Language model pre-training has been shown to be effective for improving many natural language processing tasks ( Dai and Le , 2015 ; Peters et al. , 2018a ; Radford et al. , 2018 ; Howard and Ruder , 2018 ) . These include sentence-level tasks such as natural language inference ( Bowman et al. , 2015 ; Williams et al. , 2018 ) and paraphrasing ( Dolan and Brockett , 2005 ) , which aim to predict the relationships between sentences by analyzing them holistically , as well as token-level tasks such as named entity recognition and question answering , where models are required to produce fine-grained output at the token level ( Tjong Kim Sang and De Meulder , 2003 ; Rajpurkar et al. , 2016 ) . There are two existing strategies for applying pre-trained language representations to downstream tasks : feature-based and fine-tuning . The feature-based approach , such as ELMo ( Peters et al. , 2018a ) , uses task-specific architectures that include the pre-trained representations as additional features . The fine-tuning approach , such as the Generative Pre-trained Transformer ( OpenAI GPT ) ( Radford et al. , 2018 ) , introduces minimal task-specific parameters , and is trained on the downstream tasks by simply fine-tuning all pretrained parameters . The two approaches share the same objective function during pre-training , where they use unidirectional language models to learn general language representations . We argue that current techniques restrict the power of the pre-trained representations , especially for the fine-tuning approaches . The major limitation is that standard language models are unidirectional , and this limits the choice of architectures that can be used during pre-training . For example , in OpenAI GPT , the authors use a left-toright architecture , where every token can only attend to previous tokens in the self-attention layers of the Transformer ( Vaswani et al. , 2017 ) . Such restrictions are sub-optimal for sentence-level tasks , and could be very harmful when applying finetuning based approaches to token-level tasks such as question answering , where it is crucial to incorporate context from both directions . In this paper , we improve the fine-tuning based approaches by proposing BERT : Bidirectional Encoder Representations from Transformers . BERT alleviates the previously mentioned unidirectionality constraint by using a `` masked language model '' ( MLM ) pre-training objective , inspired by the Cloze task ( Taylor , 1953 ) . The masked language model randomly masks some of the tokens from the input , and the objective is to predict the original vocabulary id of the masked arXiv:1810.04805v2 [ cs.CL ] 24 May 2019 word based only on its context . Unlike left-toright language model pre-training , the MLM objective enables the representation to fuse the left and the right context , which allows us to pretrain a deep bidirectional Transformer . In addition to the masked language model , we also use a `` next sentence prediction '' task that jointly pretrains text-pair representations . The contributions of our paper are as follows : • We demonstrate the importance of bidirectional pre-training for language representations . Unlike Radford et al . ( 2018 ) , which uses unidirectional language models for pre-training , BERT uses masked language models to enable pretrained deep bidirectional representations . This is also in contrast to Peters et al . ( 2018a ) , which uses a shallow concatenation of independently trained left-to-right and right-to-left LMs . • We show that pre-trained representations reduce the need for many heavily-engineered taskspecific architectures . BERT is the first finetuning based representation model that achieves state-of-the-art performance on a large suite of sentence-level and token-level tasks , outperforming many task-specific architectures . • BERT advances the state of the art for eleven NLP tasks . The code and pre-trained models are available at https : //github.com/ google-research/bert . "