| --- |
| title: README |
| emoji: 🔴 |
| colorFrom: red |
| colorTo: gray |
| sdk: static |
| pinned: false |
| --- |
| |
| <p align="center"> |
| <img src="https://huggingface.co/spaces/epfl-dlab/README/resolve/main/assets/dlab_header.png" alt="EPFL · dlab" width="330"> |
| </p> |
|
|
| <h1 align="center">EPFL Data Science & AI Lab (dlab)</h1> |
|
|
| <p align="center"><em><strong>AI safety</strong> · alignment · language models</em></p> |
|
|
| The **Data Science & AI Lab** at [EPFL](https://www.epfl.ch), led by [Robert West](https://dlab.epfl.ch/people/west/), |
| works on **AI safety and alignment**: how language models acquire values, how those values can |
| be installed early enough to hold, and how to tell whether they actually did. Alongside that we |
| study what models represent internally and how to make them more efficient. This Hub hosts the |
| models, datasets, and benchmarks from our public releases. |
|
|
| ## Research projects |
|
|
| | Project | Paper | Code | On the Hub | |
| | --- | :---: | :---: | :---: | |
| | **SPP** — installing an assistant persona from token zero | — | [GitHub](https://github.com/epfl-dlab/model-raising-data) | [🏛️ dlab-spp](https://huggingface.co/dlab-spp) | |
| | **zip2zip** — inference-time adaptive tokenization via online compression | [arXiv](https://arxiv.org/abs/2506.01084) | [GitHub](https://github.com/epfl-dlab/zip2zip) | [Models](https://huggingface.co/collections/epfl-dlab/zip2zip-models-6852ec90f3dacc02aa6a0dca) | |
| | **JSONSchemaBench** — benchmarking constrained decoding on real-world JSON schemas | [arXiv](https://arxiv.org/abs/2501.10868) | [GitHub](https://github.com/guidance-ai/jsonschemabench) | [Dataset](https://huggingface.co/datasets/epfl-dlab/JSONSchemaBench) | |
| | **Llaza** — pretraining data mixtures | — | — | [Collection](https://huggingface.co/collections/epfl-dlab/llaza-ft-v01-69f7bb69776cea9530672480) | |
|
|
| **Synthetic Persona Pretraining** has its own organization — [**@dlab-spp**](https://huggingface.co/dlab-spp) — |
| with all models, the pretraining and post-training data, and the evaluation benchmarks. |
|
|
| Browse [all GitHub repositories](https://github.com/epfl-dlab) or visit the [lab website](https://dlab.epfl.ch). |
|
|
