3 citations · 5 across the 33 of their papers we have counts for
39 papers
KV-Kaizen: Learning Context-Adaptive Cache Compression Choices
Joao Monteiro, Louis Béthune, Anastasiia Filippova +3
As the context size of text processed with an LLM grows, the size of KV caches can outstrip the memory allocated for the original model weights. This impacts LLM throughput negativ…
Cartridges++: KV Cache Compression without Off-Context Derailment
Sonia Laguna, Joao Monteiro, Marco Cuturi +2
Serving long documents to a Large Language Model (LLM) repeatedly is expensive: computations grow with context length, and the memory footprint of the key-value (KV) cache balloons…
KV-Lingo: Learning KV-Cache Translators with Distillation
Valérie Castin, Keitaro Sakamoto, Anastasiia Filippova +3
Large language models represent context with a key-value (KV) cache. Caches are model-specific: for the same text, models with different architectures or weights produce incompatib…
Brenier Meets Adversarial Training: Optimal Transport Geometry for Robust Learning
Alireza Abdollahpoorrostam, Ehsan Sharifian, Buse Şen +2
Distributionally robust optimization (DRO) provides a principled framework for learning under distribution shift, but its practical use is hindered by the difficulty of evaluating…
Locking Pretrained Weights via Deep Low-Rank Residual Distillation
Keitaro Sakamoto, Pierre Ablin, Federico Danieli +1
The quality of open-weight language models has dramatically improved in recent years. Sharing weights greatly facilitates model adoption by enabling their use across diverse hardwa…
DynaMiCS: Fine-tuning LLMs with Performance Constraints using Dynamic Mixtures
Eleonora Gualdoni, Sonia Laguna, Louis Bethune +3
Multi-domain fine-tuning of large language models requires improving performance on target domains while preserving previous capabilities, such as general knowledge, instruction fo…
Scaling Categorical Flow Maps
Oscar Davis, Anastasiia Filippova, Pierre Ablin +4
Continuous diffusion and flow matching models could represent a powerful alternative to autoregressive approaches for language modelling (LM), as they unlock a host of advantages c…
HyperTransport: Amortized Conditioning of T2I Generative Models
Valentino Maiorca, Eleonora Gualdoni, Xavier Suau +3
As foundation models grow in capability, the ability to efficiently and reliably control their behavior becomes critical. Fine-tuning these models can be costly, and while promptin…
Nectar: Neural Estimation of Cached-Token Attention via Regression
João Monteiro, Michal Klein, Pierre Ablin +1
Evaluating softmax attention over a fixed long context requires reading every cached key-value pair for each new query token. For a given context (a book, a manual, a legal corpus)…
Stochastic KV Routing: Enabling Adaptive Depth-Wise Cache Sharing
Anastasiia Filippova, David Grangier, Marco Cuturi +1
Serving transformer language models with high throughput requires caching Key-Values (KVs) to avoid redundant computation during autoregressive generation. The memory footprint of…
The Coupling Within: Flow Matching via Distilled Normalizing Flows
David Berthelot, Tianrong Chen, Jiatao Gu +6
Flow models have rapidly become the go-to method for training and deploying large-scale generators, owing their success to inference-time flexibility via adjustable integration ste…
Amortized Maximum Inner Product Search with Learned Support Functions
Theo X. Olausson, João Monteiro, Michal Klein +1
Maximum inner product search (MIPS) is a crucial subroutine in machine learning, requiring the identification of a vector taken within a database (the keys) that best aligns with a…
The Design Space of Tri-Modal Masked Diffusion Models
Louis Bethune, Victor Turrisi, Bruno Kacper Mlodozeniec +21
Discrete diffusion models have emerged as strong alternatives to autoregressive language models, with recent work initializing and fine-tuning a base unimodal model for bimodal gen…
LaCy: What Small Language Models Can and Should Learn is Not Just a Question of Loss
Szilvia Ujváry, Louis Béthune, Pierre Ablin +3
Language models have consistently grown to compress more world knowledge into their parameters, but the knowledge that can be pretrained into them is upper-bounded by their paramet…
Completed Hyperparameter Transfer across Modules, Width, Depth, Batch and Duration
Bruno Mlodozeniec, Pierre Ablin, Louis Béthune +4
Hyperparameter tuning can dramatically impact training stability and final performance of large-scale models. Recent works on neural network parameterisations, such as P, have e…
Learning Unmasking Policies for Diffusion Language Models
Metod Jazbec, Theo X. Olausson, Louis Béthune +6
Diffusion (Large) Language Models (dLLMs) now match the downstream performance of their autoregressive counterparts on many tasks, while holding the promise of being more efficient…
Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretraining
Thiziri Nait Saada, Louis Bethune, Michal Klein +3
Large-scale models are pretrained on massive web-crawled datasets containing documents of mixed quality, making data filtering essential. A popular method is Classifier-based Quali…
Flow Matching with Semidiscrete Couplings
Alireza Mousavi-Hosseini, Stephen Y. Zhang, Michal Klein +1
Flow models parameterized as time-dependent velocity fields can generate data from noise by integrating an ODE. These models are often trained using flow matching, i.e. by sampling…
The Geometries of Truth Are Orthogonal Across Tasks
Waiss Azizian, Michael Kirchhof, Eugene Ndiaye +4
Large Language Models (LLMs) have demonstrated impressive generalization capabilities across various tasks, but their claim to practical relevance is still mired by concerns on the…
On Fitting Flow Models with Large Sinkhorn Couplings
Stephen Zhang, Alireza Mousavi-Hosseini, Michal Klein +1
Flow models transform data gradually from one modality (e.g. noise) onto another (e.g. images). Such models are parameterized by a time-dependent velocity field, trained to fit seg…
LinEAS: End-to-end Learning of Activation Steering with a Distributional Loss
Pau Rodriguez, Michal Klein, Eleonora Gualdoni +5
The growing use of generative models in daily life calls for efficient mechanisms to control their generation, to e.g., produce safe content or provide users with tools to explore…
Sample and Map from a Single Convex Potential: Generation using Conjugate Moment Measures
Nina Vesseron, Louis Béthune, Marco Cuturi
The canonical approach in generative modeling is to split model fitting into two blocks: define first how to sample noise (e.g. Gaussian) and choose next what to do with it (e.g. u…
Multivariate Conformal Prediction using Optimal Transport
Michal Klein, Louis Bethune, Eugene Ndiaye +1
Conformal prediction (CP) quantifies the uncertainty of machine learning models by constructing sets of plausible outputs. These sets are constructed by leveraging a so-called conf…
Scaling Laws for Forgetting during Finetuning with Pretraining Data Injection
Louis Bethune, David Grangier, Dan Busbridge +3
A widespread strategy to obtain a language model that performs well on a target domain is to finetune a pretrained model to perform unsupervised next-token prediction on data from…