A temporary CLM phase followed by MLM decay during encoder continued pretraining outperforms standard MLM on biomedical tasks by 0.3-2.8pp across languages and model sizes.
needle”) is inserted into a clinical document (the “haystack
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Causal Language Modeling Detour Improves Encoder Continued Pretraining
A temporary CLM phase followed by MLM decay during encoder continued pretraining outperforms standard MLM on biomedical tasks by 0.3-2.8pp across languages and model sizes.