Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · facebook

bart-base

bart-base is a 139M-parameter BART encoder-decoder model from facebook. According to the model card, it is the base-sized English model meant chiefly for fine-tuning.

Publisher
facebook
Task
feature-extraction
Model type
bart
License
apache-2.0
Library
transformers
Publication status
Accepted · not indexed

Model overview

bart-base is published by facebook as a bart feature-extraction model. The captured configuration identifies BartModel with model type bart, and Safetensors metadata reports 139,420,416 parameters. According to the model card, it is the base-sized English BART encoder-decoder introduced by Lewis et al., with card data recording apache-2.0.

Recorded capabilities

Encoder-decoder denoising model

According to the model card, BART combines a bidirectional encoder with an autoregressive decoder and is pretrained by corrupting text and learning to reconstruct it.

Fine-tuning-oriented base model

According to the model card, BART is effective when fine-tuned for text generation such as summarization and translation, and also works for classification and question answering.

Compact Transformers record

The captured configuration identifies BartModel with model type bart and about 139M Safetensors parameters with Transformers, PyTorch, TensorFlow, and JAX tags.

Use cases in the source record

  • Fine-tuning for summarization or translation, which the model card identifies as effective uses of BART.
  • Fine-tuning for comprehension tasks such as text classification or question answering.
  • Raw-model text infilling, which the model card notes as a direct use before task-specific fine-tuning.

Limitations and unknowns

  • According to the model card, the publisher did not write a card for this model; the captured card was written by the Hugging Face team.
  • No evaluation results, context-window value, or training compute details were extracted.
  • Provider state is historical snapshot data and should be refreshed before being presented as current.

Source and provenance

Source: facebook/bart-base

Captured: Unknown. Processed: 2026-09-07T19:34:44.261441+00:00.

BART (base-sized model) BART model pre-trained on English language. It was introduced in the paper BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension by Lewis et al. and first released in this repository . Disclaimer: The team releasing BART did not write a model card for this model so this model card has been written by the Hugging Face team. Model description BART is a transformer encoder-decoder (seq2seq) model with a bidirectional (BERT-like) encoder and an autoregressive (GPT-like) decoder. BART is pre-trained by (1) corrupting text with an arbitrary noising functio…

F001F002F003F004F005F006F007F008F010F011F012F013F014F015