9 results

sub-1b

inall models datasets apps agents people projects

Datasets

All datasets matching “sub-1b”

sub1b /babylm-didactic-blend-corpora Datasheet: didactic-blend custom training corpora (BabyLM 2026 Strict-Small) Download: https://huggingface.co/datasets/sub1b/babylm-didactic-blend-corpora (model + checkpoint ladder: https://huggingface.co/sub1b/gpt-bert-a2-strict-small) Composition: BabyLM 2026 official 10M natural corpus (unchanged; redistributed per challenge terms) blended per arm with: LLM-generated didactic text (competition pool 3.02M words final; d1conc 5.72M; d1spread 5.85M; d1social 9.57M)… See the full description on the dataset page: https://huggingface.co/datasets/sub1b/babylm-didactic-blend-corpora.0 likes5 downloads2mo agoHugging Face