Zobrazeno 1 - 10
of 161
pro vyhledávání: '"Gallego, Víctor"'
Autor:
Gallego, Victor
The robustness of large language models (LLMs) against adversarial manipulations, such as jailbreak attacks, remains a significant challenge. In this work, we propose an approach that enhances the self-critique capability of the LLM and further fine-
Externí odkaz:
http://arxiv.org/abs/2406.07188
Autor:
Gallego, Victor
State-of-the-art language model fine-tuning techniques, such as Direct Preference Optimization (DPO), restrict user control by hard-coding predefined behaviors into the model. To address this, we propose a novel method, Configurable Safety Tuning (CS
Externí odkaz:
http://arxiv.org/abs/2404.00495
Autor:
Gallego, Víctor
In this paper, we introduce \emph{refined Direct Preference Optimization} (rDPO), a method for improving the behavioral alignment of Large Language Models (LLMs) without the need for human-annotated data. The method involves creating synthetic data u
Externí odkaz:
http://arxiv.org/abs/2402.08005
Autor:
Gallego, Victor
Publikováno v:
The Second Tiny Papers Track at ICLR 2024
This paper proposes an interpretation of RLAIF as Bayesian inference by introducing distilled Self-Critique (dSC), which refines the outputs of a LLM through a Gibbs sampler that is later distilled into a fine-tuned model. Only requiring synthetic da
Externí odkaz:
http://arxiv.org/abs/2312.01957
Autor:
Gallego, Victor
In this work, we address the problem of directing the text generation of a language model (LM) towards a desired behavior, aligning the generated text with the preferences of the human operator. We propose using another, instruction-tuned language mo
Externí odkaz:
http://arxiv.org/abs/2308.06385
Autor:
Gallego, Victor
Recently, large multimodal models, such as CLIP and Stable Diffusion have experimented tremendous successes in both foundations and applications. However, as these models increase in parameter size and computational requirements, it becomes more chal
Externí odkaz:
http://arxiv.org/abs/2308.07929
Autor:
Gallego, Victor
This work proposes aesthetic gradients, a method to personalize a CLIP-conditioned diffusion model by guiding the generative process towards custom aesthetics defined by the user from a set of images. The approach is validated with qualitative and qu
Externí odkaz:
http://arxiv.org/abs/2209.12330
How do tuna schools associate to dFADs? A study using echo-sounder buoys to identify global patterns
Autor:
Navarro-García, Manuel, Precioso, Daniel, Gavira-O'Neill, Kathryn, Torres-Barrán, Alberto, Gordo, David, Gallego, Víctor, Gómez-Ullate, David
Based on the data gathered by echo-sounder buoys attached to drifting Fish Aggregating Devices (dFADs) across tropical oceans, the current study applies a Machine Learning protocol to examine the temporal trends of tuna schools' association to drifti
Externí odkaz:
http://arxiv.org/abs/2207.07049
Autor:
Gallego, Víctor
The rampant adoption of ML methodologies has revealed that models are usually adopted to make decisions without taking into account the uncertainties in their predictions. More critically, they can be vulnerable to adversarial examples. Thus, we beli
Externí odkaz:
http://arxiv.org/abs/2109.13232
Data sharing issues pervade online social and economic environments. To foster social progress, it is important to develop models of the interaction between data producers and consumers that can promote the rise of cooperation between the involved pa
Externí odkaz:
http://arxiv.org/abs/2101.10721