Salamandra Technical Report

TL;DR

Salamandra is an open-source suite of multilingual, decoder-only large language models trained on diverse European languages and code, with supplementary fine-tuned chat models and initial multimodal experiments, promoting transparency and open science.

Contribution

This work introduces Salamandra, a new open-source multilingual LLM suite with detailed training, evaluation, and open access, including models, scripts, and preliminary multimodal capabilities.

Findings

Salamandra models achieve competitive performance on multilingual benchmarks.

Extensive evaluations cover downstream tasks, bias, and safety.

Open-source release fosters research and commercial use.

Abstract

This work introduces Salamandra, a suite of open-source decoder-only large language models available in three different sizes: 2, 7, and 40 billion parameters. The models were trained from scratch on highly multilingual data that comprises text in 35 European languages and code. Our carefully curated corpus is made exclusively from open-access data compiled from a wide variety of sources. Along with the base models, supplementary checkpoints that were fine-tuned on public-domain instruction data are also released for chat applications. Additionally, we also share our preliminary experiments on multimodality, which serve as proof-of-concept to showcase potential applications for the Salamandra family. Our extensive evaluations on multilingual benchmarks reveal that Salamandra has strong capabilities, achieving competitive performance when compared to similarly sized open-source models. We provide comprehensive evaluation results both on standard downstream tasks as well as key aspects related to bias and safety.With this technical report, we intend to promote open science by sharing all the details behind our design choices, data curation strategy and evaluation methodology. In addition to that, we deviate from the usual practice by making our training and evaluation scripts publicly accessible. We release all models under a permissive Apache 2.0 license in order to foster future research and facilitate commercial use, thereby contributing to the open-source ecosystem of large language models.