pheinisch's picture
Update README.md
e1f5dde
|
raw
history blame
671 Bytes
metadata
language:
  - en
tags:
  - argument mining
  - claims
  - sentence classification
datasets:
  - FS150T
metrics:
  - accuracy
  - f1

EXPERIMENTAL roberta-base-150T-argumentative-sentence-detector

(this model might not be the optimal one for accomplishing the task)

  • Task: Detects whether a sentence is argumentative (1 - yes/ 0 - not) given the topic and the sentence itself.
  • language: English
  • dataset: Few-Shot-150T Corpus v1.1 (FS150T-Corpus) fin-tuned roberta-base

Performace on test data (threshold: 0.5)

{'accuracy': 0.7451388888888889,
 'f1': 0.6690712353471596,
 'precision': 0.733201581027668,
 'recall': 0.615257048092869}