ARAML stabilizes adversarial text generation by training the generator with reward-weighted maximum likelihood on samples drawn from a fixed distribution around real data instead of using policy gradient.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
ARAML: A Stable Adversarial Training Framework for Text Generation
ARAML stabilizes adversarial text generation by training the generator with reward-weighted maximum likelihood on samples drawn from a fixed distribution around real data instead of using policy gradient.