--- title: README emoji: 🧭 colorFrom: green colorTo: indigo sdk: static pinned: false ---

SPP: annotate pretraining data with normative reflections, inject at different pretraining stages, evaluate alignment and safety

Alignment — and the assistant identity itself — is normally introduced only *after* pretraining, once behavioral priors are already set. **SPP** installs the desired persona from token zero instead: we define it through normative values in a constitution, generate first-person moral reflections grounded in that constitution, and insert them throughout the pretraining corpus behind an `` token. Post-training then binds the chat assistant identity to the installed persona. Pretraining up to **3B on 500B tokens**, SPP improves constitution following and jailbreak robustness while preserving capabilities — and *when* the data arrives matters: models trained with reflections from token zero prioritize values differently and take fewer risky actions in out-of-distribution moral dilemmas than models given the exact same data only at the end of pretraining, an advantage that **grows with scale**. ## Collections 📦 **[Pretraining Datasets](https://huggingface.co/collections/dlab-spp/pretraining-datasets-6a6b3b62c372179e342a1de6)** — the reflection data, the corpus selection manifest, safety scores, and verification files. 🤖 **[Models — 3B](https://huggingface.co/collections/dlab-spp/models-3b-6a6b3cbc2bf4be9aba1c4a16)** · **[Models — 1.7B](https://huggingface.co/collections/dlab-spp/models-17b-6a6b3cbf48d091f505d412aa)** — all five recipes, at both scales. We release all pretraining checkpoints, base, and instruct models, at both scales 💬 **[Post-training Dataset](https://huggingface.co/collections/dlab-spp/post-training-dataset-6a6c801cd9d46e4ff5d2ac25)** — SP-SFT, the mixture that performs persona binding. 📊 **[Evals](https://huggingface.co/collections/dlab-spp/evals-6a6c810619bb9009709286da)** — ConstitutionEval and an audited AIRiskDilemmas. _{From [EPFL DLAB](https://huggingface.co/epfl-dlab).}