---
license: other
language:
- en
library_name: transformers
pipeline_tag: text-generation
tags:
- model-raising
- synthetic-persona-pretraining
- spp
- alignment
- safety
---

# SPP-T0 — Base (3B)

**Type:** base (pretrained) model. Not instruction-tuned and ships no chat template.

Trained with Synthetic Persona Pretraining (SPP) from token zero: first-person reflections are inserted into the roughly 10% of annotated documents that carry one, throughout the entire pretraining run.

## Synthetic Persona Pretraining (SPP)
**Synthetic Persona Pretraining (SPP)** installs a target value persona during pretraining rather than only during alignment. Value-laden, first-person reflections, generated against a constitution, are appended to a subset of pretraining documents after a special `<assistant>` token. Attention masking and RoPE position aliasing keep the reflection from changing the continuation of the original document. This model is trained with SPP.

Instruction-tuned counterpart: [`epfl-dlab/spp-t0-3b-instruct`](https://huggingface.co/epfl-dlab/spp-t0-3b-instruct).

## Model details
- **Architecture:** Llama-3.2-3B-shaped, trained from scratch.
- **Tokenizer:** the SmolLM2 tokenizer extended with an `<assistant>` marker and constitution tokens (vocabulary 49280).
- **Pretraining:** ~500B tokens on a subset of the Olmo 3 Dolma 3 mixture, with SPP reflections inserted into the safety-annotated documents within it.

## Training checkpoints

Intermediate checkpoints are published as git revisions on this repo, so any point in the trajectory can be loaded by passing `revision=`:

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

repo = "epfl-dlab/spp-t0-3b-base"
tok = AutoTokenizer.from_pretrained(repo)          # identical at every revision
model = AutoModelForCausalLM.from_pretrained(
    repo, revision="step-25000", dtype=torch.bfloat16, device_map="auto"
)
```

| Revision | Pretraining step | Tokens seen | LR phase |
|---|---|---|---|
| `step-25000` | 25,000 / 254,313 | ~49.2B | stable |
| `step-50000` | 50,000 / 254,313 | ~98.3B | stable |
| `step-75000` | 75,000 / 254,313 | ~147B | stable |
| `step-100000` | 100,000 / 254,313 | ~197B | stable |
| `step-125000` | 125,000 / 254,313 | ~246B | stable |
| `step-150000` | 150,000 / 254,313 | ~295B | stable |
| `step-175000` | 175,000 / 254,313 | ~344B | stable |
| `step-200000` | 200,000 / 254,313 | ~393B | stable |
| `step-225000` | 225,000 / 254,313 | ~442B | stable |
| `step-240000` | 240,000 / 254,313 | ~472B | linear decay |
| `step-254313` | 254,313 / 254,313 | ~500B | linear decay — same weights as `main` |

`main` always holds the finished model (step 254,313).
Only model weights are published — optimizer and RNG state are not included, so these revisions support evaluation, probing, and fine-tuning, but not exact resumption of the original run.

## Intended use
Research on alignment and safety. As a base model it is meant for continuation, probing, or further fine-tuning; it is not instruction-tuned and can produce incorrect or unsafe content.

## Links
- Paper: _to be released_
- Collection: https://huggingface.co/collections/epfl-dlab/spp-synthetic-persona-pretraining

_License: to be finalised._