Concept

Écriture bicamérale

Publications associées (146)

Beyond fine-tuning: LoRA modules boost near-OOD detection and LLM security

Pierre Vandergheynst, Milos Vasic, Francesco Craighero, Renata Khasanova

Under resource constraints, LLMs are usually fine- tuned with additional knowledge using Parameter Efficient Fine-Tuning (PEFT), using Low-Rank Adaptation (LoRA) modules. In fact, LoRA injects a new set of small trainable matrices to adapt an LLM to a new ...

2024

Beyond fine-tuning: LoRA modules boost near-OOD detection and LLM security

Pierre Vandergheynst, Milos Vasic, Francesco Craighero, Renata Khasanova

Under resource constraints, LLMs are usually fine-tuned with additional knowledge using Parameter Efficient Fine-Tuning (PEFT), using Low-Rank Adaptation (LoRA) modules. In fact, LoRA injects a new set of small trainable matrices to adapt an LLM to a new t ...

2024

EAST discharge prediction without integrating simulation results (vol 62, 126060, 2022)

Alessandro Pau, Xiaojuan Liu, Chenguang Wan

In the published paper titled 'EAST discharge prediction without integrating simulation results' (2022 Nucl. Fusion 62 126060), figure 5 used the wrong plot, and is not matched with table 3. We used the plot discussed with referees, the case when we do not ...

Bristol2023

Noncollinear DFT plus U and Hubbard parameters with fully relativistic ultrasoft pseudopotentials

Nicola Marzari, Luca Binci

The magnetic, noncollinear parametrization of Dudarev's DFT + U method is generalized to fully relativistic ultrasoft pseudopotentials. We present the definition of the DFT + U total energy functional and the calculation of forces and stresses in the case ...

College Pk2023

Alternation makes the adversary weaker in two-player games

Volkan Cevher, Efstratios Panteleimon Skoulakis, Luca Viano, Ali Kavis

Motivated by alternating game-play in two-player games, we study an altenating variant of the Online Linear Optimization (OLO). In alternating OLO, a learner at each round t ∈[n] selects a vector xt and then an adversary selects a cost-vector ct ∈[−1,1]n. ...

2023

A tale of two coal regimes: An actor-oriented analysis of destabilisation and maintenance of coal regimes in Germany and Japan

Mert Duygan

Phasing out coal-fired electricity is an urgent global task, critical to efforts to mitigate climate change and air pollution. Despite the growth and increasing competitiveness of renewable energy, phase-out progress is slow, with coal-fired power even rea ...

Amsterdam2023

Lie groups in the symmetric group: Reducing Ulam's problem to the simple case

Nicolas Monod

Ulam asked whether all Lie groups can be represented faithfully on a countable set. We establish a reduction of Ulam's problem to the case of simple Lie groups. In particular, we solve the problem for all solvable Lie groups and more generally Lie groups w ...

San Diego2023

Optima Age Over Erasure Channels

Emre Telatar, Elie Najm, Rajai Nasser

Previous works on age of information and erasure channels have dealt with specific models and computed the average age or average peak age for certain settings. In this paper, given a source that produces a letter every T-s seconds and an erasure channel t ...

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC2022

A new insight into the dynamic impact between geophysical flow and rigid barrier

Xingyue Li

Understanding the interaction between complicated geophysical flows and barriers remains a critical challenge for hazards countermeasures in engineering design. In this study, the impact of geophysical flows on a rigid barrier is studied numerically using ...

ELSEVIER SCI LTD2022

The Price of Distributed: Rate Loss in the CEO Problem

Michael Christoph Gastpar, Alper Köse, Ahmet Arda Atalik

In the distributed remote (CEO) source coding problem, many separate encoders observe independently noisy copies of an underlying source. The rate loss is the difference between the rate required in this distributed setting and the rate that would be requi ...

2022

Graph Chatbot

Chattez avec Graph Search

Posez n’importe quelle question sur les cours, conférences, exercices, recherches, actualités, etc. de l’EPFL ou essayez les exemples de questions ci-dessous.

AVERTISSEMENT : Le chatbot Graph n'est pas programmé pour fournir des réponses explicites ou catégoriques à vos questions. Il transforme plutôt vos questions en demandes API qui sont distribuées aux différents services informatiques officiellement administrés par l'EPFL. Son but est uniquement de collecter et de recommander des références pertinentes à des contenus que vous pouvez explorer pour vous aider à répondre à vos questions.

Connectez-vous pour utiliser Chat avec Graph Search