A 564M-parameter model trained on teacher-generated NER and QA annotations over generic text matches or beats a 13B teacher on some benchmarks, though the comparison is not always clean.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Basic Reading Distillation
A 564M-parameter model trained on teacher-generated NER and QA annotations over generic text matches or beats a 13B teacher on some benchmarks, though the comparison is not always clean.