The method used to train NLP systems by masking or substituting some input words belongs to the category of contrastive methods.
Not checkable An opinion, value judgement or statement with nothing to check.
Not checkable: An opinion, value judgement or statement with nothing to check. Not scored.
Cite this claim
Yann LeCun, “The method used to train NLP systems by masking or substituting some input words belongs to the category of contrastive methods.” (said 4 March 2021). Verdict: not checkable. Reliability API, claim c_e7b3688b50c4, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_e7b3688b50c4
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
Verdicts are made automatically; how verdicts are made.