2 months ago

Pixtral 12B

{Sophia Yang, Baptiste Rozière, Nikhil Raghuraman, Patrick von Platen, Marie Pellat, Valera Nemychnikova, Pavankumar Muddireddy, Arthur Mensch, Louis Martin, William Marshall, Andy Lo, Teven Le Scao, Thibaut Lavril, Diego Las Casas, Guillaume Lample, Timothée Lacroix, Kartik Khandelwal, Albert Q. Jiang, Paul Jacob, Amélie Héliou, Baudouin De Monicault, Diogo Costa, Baptiste Bout, Thomas Wang, Sagar Vaze, Sandeep Subramanian, Joachim Studnia, Pierre Stock, Lawrence Stewart, Roman Soletskyi, Wendy Shang, Romain Sauvestre, Lucile Saulnier, Alexandre Sablayrolles, Soham Ghosh, Theophile Gervet, Saurabh Garg, Jessica Chudnovsky, Devendra Chaplot, Emma Bou Hanna, Szymon Antoniak, Pravesh Agrawal}

View Paper Details View Code

Abstract

We introduce Pixtral-12B, a 12--billion-parameter multimodal language model. Pixtral-12B is trained to understand both natural images and documents, achieving leading performance on various multimodal benchmarks, surpassing a number of larger models. Unlike many open-source models, Pixtral is also a cutting-edge text model for its size, and does not compromise on natural language performance to excel in multimodal tasks. Pixtral uses a new vision encoder trained from scratch, which allows it to ingest images at their natural resolution and aspect ratio. This gives users flexibility on the number of tokens used to process an image. Pixtral is also able to process any number of images in its long context window of 128K tokens. Pixtral 12B substanially outperforms other open models of similar sizes (Llama-3.2 11B & Qwen-2-VL 7B). It also outperforms much larger open models like Llama-3.2 90B while being 7x smaller. We further contribute an open-source benchmark, MM-MT-Bench, for evaluating vision-language models in practical scenarios, and provide detailed analysis and code for standardized evaluation protocols for multimodal LLMs. Pixtral-12B is released under Apache 2.0 license.