3B SPP models: bases publish their training trajectory as step-* revisions, plus persona-binding instruct counterparts.
AI & ML interests
None defined yet.
Recent Activity
The SPP pretraining corpus: reflections, the corpus selection manifest, safety scores, and verification data. Fully reproducible from public Dolma 3.
1.7B SPP models: bases publish their training trajectory as step-* revisions, plus persona-binding instruct counterparts.
3B SPP models: bases publish their training trajectory as step-* revisions, plus persona-binding instruct counterparts.
SP-SFT: the post-training mix that binds the assistant identity to the persona installed during pretraining.
The SPP pretraining corpus: reflections, the corpus selection manifest, safety scores, and verification data. Fully reproducible from public Dolma 3.
ConstitutionEval (in-domain constitution following) and audited AIRiskDilemmas (out-of-distribution value prioritisation and risk-taking).
1.7B SPP models: bases publish their training trajectory as step-* revisions, plus persona-binding instruct counterparts.