papers

Publications (9)

cs.CL2026

Ministral 3

Alexander H. Liu, Kartik Khandelwal, Sandeep Subramanian +116

We introduce the Ministral 3 series, a family of parameter-efficient dense language models designed for compute and memory constrained applications, available in three model sizes:…

cs.CL2026

Shieldstral

Antonia Calvi, Avinash Sooriyarachchi, Giada Pistilli +274

We introduce Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier that matches or outperforms models nearly 7 its size on text safety benchmarks and set…

cs.LG2025

Residual Stream Analysis with Multi-Layer SAEs

Tim Lawson, Lucy Farnik, Conor Houghton +1

Sparse autoencoders (SAEs) are a promising approach to interpreting the internal representations of transformer language models. However, SAEs are usually trained separately on eac…

cs.AI2026

Voxtral Realtime

Mistral-AI, :, Alexander H. Liu +166

We introduce Voxtral Realtime, a natively streaming automatic speech recognition model that matches offline transcription quality at sub-second latency. Unlike approaches that adap…

cs.LG2025

Learning to Skip the Middle Layers of Transformers

Tim Lawson, Laurence Aitchison

Conditional computation is a popular strategy to make Transformers more efficient. Existing methods often target individual modules (e.g., mixture-of-experts layers) or skip layers…

cs.LG2026

Automated Interpretability Metrics Do Not Distinguish Trained and Random Transformers

Thomas Heap, Tim Lawson, Lucy Farnik +1

Sparse autoencoders (SAEs) are widely used to extract sparse, interpretable latents from transformer activations. We test whether commonly used SAE quality metrics and automatic ex…

cs.LG2025

Jacobian Sparse Autoencoders: Sparsify Computations, Not Just Activations

Lucy Farnik, Tim Lawson, Conor Houghton +1

Sparse autoencoders (SAEs) have been successfully used to discover sparse and human-interpretable representations of the latent activations of LLMs. However, we would ultimately li…

cs.RO2026

Robostral Navigate

Abhijeet Somani, Arjun Majumdar, Abdelaziz Bounhar +273

Deploying navigation systems at scale requires a recipe that minimizes sensor assumptions, generalizes across robot embodiments, and trains efficiently. Yet, today's best systems d…

cs.AI2026

Voxtral TTS

Mistral-AI, :, Alexander H. Liu +186

We introduce Voxtral TTS, an expressive multilingual text-to-speech model that generates natural speech from as little as 3 seconds of reference audio. Voxtral TTS adopts a hybrid…