Small sanity corpus of standard rank-16 LoRAs trained on 25 organisms sampled from
ceselder/loracle-pretrain.
Each organism is a set of embedded source documents; no external FineFineWeb pull is required.
loras/
.pt: raw MultiTaskLoRA weight dicts, not PEFT adapter directories.
direction_tokens_svd_k16/.pt: residual-side SVD direction tokens.
selected_organisms.parquet: one row per organism, including… See the full description on the dataset page:
https://huggingface.co/datasets/cds-jb/loracles-qwen3-8b-pretrain-loras.
