Vincent Gripon

Sharing LoRA adapters between teacher and student models unlocks faster and more effective knowledge distillation, boosting performance for both.

Discrete diffusion models can now generate more diverse text without sacrificing quality, thanks to a new decoding method that explicitly optimizes for diversity during beam search.

MONET reveals the potential for significant hardware architecture improvements by modeling and optimizing neural network training, a domain often overshadowed by inference-centric design.

Unlock hidden performance in your pre-trained language models with "inner looping," a simple inference-time trick that repeatedly refines latent representations by re-applying selected transformer blocks.

Transformers suffer from a subtle but significant misalignment: residual connections inadvertently tie information to the *wrong* token, but a simple residual attenuation fix can boost performance.

Skip the training: MUKA, a multi-kernel adaptation framework, lets Large Audio-Language Models achieve state-of-the-art few-shot performance without any additional training.