ccdv's picture
small fix
e1c052c
metadata
tags:
  - summarization
  - summary
  - booksum
  - long-document
  - long-form
  - lsg
datasets:
  - kmfoda/booksum
metrics:
  - rouge
model-index:
  - name: ccdv/lsg-bart-base-4096-booksum
    results: []

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.

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline

tokenizer = AutoTokenizer.from_pretrained("ccdv/lsg-bart-base-4096-booksum", trust_remote_code=True)
model = AutoModelForSeq2SeqLM.from_pretrained("ccdv/lsg-bart-base-4096-booksum", 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
  )

ccdv/lsg-bart-base-4096-booksum

This model is a fine-tuned version of ccdv/lsg-bart-base-4096 on the kmfoda/booksum kmfoda--booksum dataset. It achieves the following results on the evaluation set:

  • eval_loss: 3.2654
  • eval_rouge1: 33.9468
  • eval_rouge2: 6.7034
  • eval_rougeL: 16.7879
  • eval_rougeLsum: 31.7677
  • eval_gen_len: 427.6918
  • eval_runtime: 2910.3841
  • eval_samples_per_second: 0.492
  • eval_steps_per_second: 0.062
  • eval_samples: 1431

Model description

More information needed

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: 30.0

Generate hyperparameters

The following hyperparameters were used during generation:

  • dataset_name: kmfoda/booksum
  • dataset_config_name: kmfoda--booksum
  • eval_batch_size: 8
  • eval_samples: 1431
  • early_stopping: True
  • ignore_pad_token_for_loss: True
  • length_penalty: 2.0
  • max_length: 512
  • min_length: 128
  • num_beams: 5
  • no_repeat_ngram_size: None
  • seed: 123

Framework versions

  • Transformers 4.36.1
  • Pytorch 1.12.1
  • Datasets 2.3.2
  • Tokenizers 0.11.6