---
date: '2022-11-28T11:31:57'
hypothesis-meta:
  created: '2022-11-28T11:31:57.626263+00:00'
  document:
    title:
    - Countering the Effects of Lead Bias in News Summarization via Multi-Stage Training
      and Auxiliary Losses
  flagged: false
  group: __world__
  hidden: false
  id: RG4WUG8QEe2-tZ8v6nNOJA
  links:
    html: https://hypothes.is/a/RG4WUG8QEe2-tZ8v6nNOJA
    incontext: https://hyp.is/RG4WUG8QEe2-tZ8v6nNOJA/aclanthology.org/D19-1620.pdf
    json: https://hypothes.is/api/annotations/RG4WUG8QEe2-tZ8v6nNOJA
  permissions:
    admin:
    - acct:ravenscroftj@hypothes.is
    delete:
    - acct:ravenscroftj@hypothes.is
    read:
    - group:__world__
    update:
    - acct:ravenscroftj@hypothes.is
  tags:
  - NLProc
  - summarization
  - bandit
  - rl
  target:
  - selector:
    - end: 7922
      start: 7890
      type: TextPositionSelector
    - exact: BanditSum a hierarchical bi-LSTM
      prefix: S uses a CNN+bi-GRU encoder, and
      suffix: ". RNES\u2019s de-coder is auto-regres"
      type: TextQuoteSelector
    source: https://aclanthology.org/D19-1620.pdf
  text: Banditsum uses bi-directional LSTM encoding. It generates sentence-level representations
  updated: '2022-11-28T11:34:57.447988+00:00'
  uri: https://aclanthology.org/D19-1620.pdf
  user: acct:ravenscroftj@hypothes.is
  user_info:
    display_name: James Ravenscroft
in-reply-to: https://aclanthology.org/D19-1620.pdf
tags:
- NLProc
- summarization
- bandit
- rl
- hypothesis
type: annotation
url: /annotations/2022/11/28/1669635117

---



 <blockquote>BanditSum a hierarchical bi-LSTM</blockquote>Banditsum uses bi-directional LSTM encoding. It generates sentence-level representations