CoolFace
Datasetpublic

dougalldeepmind/2026-08-02-qwen36-synthdoc-package-mixture-10-90

Qwen3.6-27B SFT mixture — synthdoc_v2 10/90 10% difficult-advice / 90% TULU3 replay, 996,193 tokens total. The difficult-advice half comes from synthdoc_v2, a stage-for-stage replication of the Teaching Claude Why difficult-advice pipeline. Source Examples Tokens Share difficult-advice (synthdoc_v2) 58 99,847 10.02% TULU3 replay 1,402 896,346 89.98% Total 1,460 996,193 md5 16b8876b9c480ca9e0351ebe92a516f2. How the difficult-advice data was made… See the full description on the dataset page: https://huggingface.co/datasets/dougalldeepmind/2026-08-02-qwen36-synthdoc-package-mixture-10-90.

sourceHugging Faceodc-byupdated 1mo agoView on Hugging Face
0likes107downloads
3 commits on main
f79221b1mo ago

backfill training-data tags

jamie-stephenson
a7192f22mo ago

synthdoc_v2 10/90 mixture, ~1M tokens, trait-balanced

matboz
58368592mo ago

initial commit

matboz