Lsg Bart Base 16384 Arxiv

🚀 ccdv/lsg-bart-base-16384-arxiv

This model is a fine - tuned version of ccdv/lsg-bart-base-4096-arxiv on the scientific_papers arxiv dataset. It can handle 16384 long sequences and has achieved certain results on the test set.

🚀 Quick Start

Transformers >= 4.36.1 This model relies on a custom modeling file, you need to add trust_remote_code=True See #13467

LSG ArXiv paper. Github/conversion script is available at this link.

Basic Usage

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline

tokenizer = AutoTokenizer.from_pretrained("ccdv/lsg-bart-base-16384-arxiv", trust_remote_code=True)
model = AutoModelForSeq2SeqLM.from_pretrained("ccdv/lsg-bart-base-16384-arxiv", trust_remote_code=True)

text = "Replace by what you want."
pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer, device=0)
generated_text = pipe(
  text, 
  truncation=True, 
  max_length=64, 
  no_repeat_ngram_size=7,
  num_beams=2,
  early_stopping=True
  )

✨ Features

• This model is a fine - tuned version of ccdv/lsg-bart-base-4096-arxiv on the scientific_papers arxiv dataset.
• It is converted to handle 16384 long sequences and fine - tuned accordingly during 1 epoch.

📚 Documentation

Model Details

This model is a fine - tuned version of ccdv/lsg-bart-base-4096-arxiv on the scientific_papers arxiv dataset. The model is converted to handle 16384 long sequences and fine - tuned accordingly during 1 epoch. It achieves the following results on the test set:

Reference model:

Model description

The model relies on Local - Sparse - Global attention to handle long sequences:

The model has about ~145 millions parameters (6 encoder layers - 6 decoder layers). The model is warm - started from ccdv/lsg-bart-base-4096-arxiv, converted to handle long sequences (encoder only) and fine - tuned.

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 8e - 05
• train_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1.0

Generate hyperparameters

The following hyperparameters were used during generation:

• dataset_name: scientific_papers
• dataset_config_name: arxiv
• eval_batch_size: 4
• eval_samples: 6440
• early_stopping: True
• ignore_pad_token_for_loss: True
• length_penalty: 2.0
• max_length: 320
• min_length: 32
• num_beams: 5
• no_repeat_ngram_size: None
• seed: 123

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.1+cu102
• Datasets 2.1.0
• Tokenizers 0.11.6