Model Details

Model Description

Abstractive summarization models are commonly trained using maximum likelihood estimation, which assumes a deterministic (one-point) target distribution in which an ideal model will assign all the probability mass to the reference summary. This assumption may lead to performance degradation during inference, where the model needs to compare several system-generated (candidate) summaries that have deviated from the reference summary. To address this problem, we propose a novel training paradigm which assumes a non-deterministic distribution so that different candidate summaries are assigned probability mass according to their quality.

• Developed by: Yale LILY Lab

• Shared by [Optional]: Yale LILY Lab

• Model type: Text2Text Generation

• Parent Model: BART

Uses

Direct Use

This model can be used for the task of Text2Text Generation

Downstream Use [Optional]

Further analysis also shows that our model can estimate probabilities of candidate summaries that are more correlated with their level of quality.

Out-of-Scope Use

The model should not be used to intentionally create hostile or alienating environments for people.

Bias, Risks, and Limitations

Significant research has explored bias and fairness issues with language models (see, e.g., Sheng et al. (2021) and Bender et al. (2021)). Predictions generated by the model may include disturbing and harmful stereotypes across protected classes; identity characteristics; and sensitive, social, and occupational groups.

Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

Training Details

Training Data

The model creators note in the associated paper:

CNNDM4: is a large scale news dataset. Nallapati et al: we treat the news articles as the source documents and the associated highlights as the summaries. XSum5: is a highly abstractive dataset of articles from the British Broadcasting Corporation (BBC). NYT6: contains articles from the New York Times and the associated summaries

Training Procedure

Preprocessing

The model creators note in the associated paper:

We follow Kedzie et al. (2018) for data preprocessing and splitting, and use the associated archival abstracts as the summaries

Evaluation

Results

CNNDM

ROUGE-1 ROUGE-2 ROUGE-L
BART 44.16 21.28 40.90
Ours 47.78 23.55 44.57

Model Examination

The model creators note in the associated paper:

We attribute BRIO-Ctr’s superior performance to its use of the same model architecture (BART) for both candidate generation and scoring, while SimCLS uses RoBERTa as the evaluation model. As a result, BRIO-Ctr maximizes the parameter sharing between the two stages, and preserves the power of the Seq2Seq model pre-trained on the same dataset.

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Technical Specifications [optional]

Model Architecture and Objective

The model creators note in the associated paper:

Formulate summarization as a sequence-to-sequence (Seq2Seq) problem

Citation

BibTeX:

@misc{mesh-transformer-jax,
@misc{https://doi.org/10.48550/arxiv.2203.16804,
doi = {10.48550/ARXIV.2203.16804},

url = {https://arxiv.org/abs/2203.16804},

author = {Liu, Yixin and Liu, Pengfei and Radev, Dragomir and Neubig, Graham},

keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},

title = {BRIO: Bringing Order to Abstractive Summarization},


Model Card Authors [optional]

Yale LILY Lab in collaboration with Ezi Ozoani and the Hugging Face team

How to Get Started with the Model

Use the code below to get started with the model.

Click to expand
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Yale-LILY/brio-cnndm-uncased")

model = AutoModelForSeq2SeqLM.from_pretrained("Yale-LILY/brio-cnndm-uncased")