Nous utilisons des témoins pour analyser le trafic et l’utilisation de notre site web, afin de personnaliser votre expérience. Vous pouvez désactiver ces technologies à tout moment, mais cela peut restreindre certaines fonctionnalités du site. Consultez notre Politique de protection de la vie privée pour en savoir plus.
Paramètre des cookies
Vous pouvez activer et désactiver les types de cookies que vous souhaitez accepter. Cependant certains choix que vous ferez pourraient affecter les services proposés sur nos sites (ex : suggestions, annonces personnalisées, etc.).
Cookies essentiels
Ces cookies sont nécessaires au fonctionnement du site et ne peuvent être désactivés. (Toujours actif)
Cookies analyse
Acceptez-vous l'utilisation de cookies pour mesurer l'audience de nos sites ?
Multimedia Player
Acceptez-vous l'utilisation de cookies pour afficher et vous permettre de regarder les contenus vidéo hébergés par nos partenaires (YouTube, etc.) ?
Publications
Predicting Grokking Long Before it Happens: A look into the loss landscape of models which grok
In order to safely deploy Large Language Models (LLMs), they must be capable of dynamically adapting their behavior based on their level of … (voir plus)knowledge and uncertainty associated with specific topics. This adaptive behavior, which we refer to as self-restraint, is non-trivial to teach since it depends on the internal knowledge of an LLM. By default, LLMs are trained to maximize the next token likelihood which does not teach the model to modulate its answer based on its level of uncertainty. In order to learn self-restraint, we devise a simple objective that can encourage the model to produce generation that the model is confident in. To optimize this objective, we introduce ReSearch, an iterative search algorithm based on self-evaluation and self-prompting. Our method results in fewer hallucinations overall, both for known and unknown topics, as the model learns to selectively restrain itself. In addition, our method elegantly incorporates the ability to decline, when the model assesses that it cannot provide a response without a high proportion of hallucination.
In order to safely deploy Large Language Models (LLMs), they must be capable of dynamically adapting their behavior based on their level of … (voir plus)knowledge and uncertainty associated with specific topics. This adaptive behavior, which we refer to as self-restraint, is non-trivial to teach since it depends on the internal knowledge of an LLM. By default, LLMs are trained to maximize the next token likelihood which does not teach the model to modulate its answer based on its level of uncertainty. In order to learn self-restraint, we devise a simple objective that can encourage the model to produce generation that the model is confident in. To optimize this objective, we introduce ReSearch, an iterative search algorithm based on self-evaluation and self-prompting. Our method results in fewer hallucinations overall, both for known and unknown topics, as the model learns to selectively restrain itself. In addition, our method elegantly incorporates the ability to decline, when the model assesses that it cannot provide a response without a high proportion of hallucination.
Protein language models are a powerful tool for learning protein representations through pre-training on vast protein sequence datasets.
Ho… (voir plus)wever, traditional protein language models lack explicit structural supervision, despite its relevance to protein function.
To address this issue, we introduce the integration of remote homology detection to distill structural information into protein language models without requiring explicit protein structures as input.
We evaluate the impact of this structure-informed training on downstream protein function prediction tasks.
Experimental results reveal consistent improvements in function annotation accuracy for EC number and GO term prediction. Performance on mutant datasets, however, varies based on the relationship between targeted properties and protein structures. This underscores the importance of considering this relationship when applying structure-aware training to protein function prediction tasks. Code and model weights will be made available upon acceptance.
Reflecting on the limitations of a study is a crucial part of the research process. In software engineering studies, this reflection is typi… (voir plus)cally conveyed through discussions of study limitations or threats to validity. In current practice, such discussions seldom provide sufficient insight to understand the rationale for decisions taken before and during the study, and their implications. We revisit the practice of discussing study limitations and threats to validity and identify its weaknesses. We propose to refocus this practice of self-reflection to a discussion centered on the notion of trade-offs. We argue that documenting trade-offs allows researchers to clarify how the benefits of their study design decisions outweigh the costs of possible alternatives. We present guidelines for reporting trade-offs in a way that promotes a fair and dispassionate assessment of researchers’ work.
2024-03-02
ACM Transactions on Software Engineering and Methodology (publié)
Ensembling multiple models enhances predictive performance by utilizing the varied learned features of the different models but incurs signi… (voir plus)ficant computational and storage costs. Model fusion, which combines parameters from multiple models into one, aims to mitigate these costs but faces practical challenges due to the complex, non-convex nature of neural network loss landscapes, where learned minima are often separated by high loss barriers. Recent works have explored using permutations to align network features, reducing the loss barrier in parameter space. However, permutations are restrictive since they assume a one-to-one mapping between the different models' neurons exists. We propose a new model merging algorithm, CCA Merge, which is based on Canonical Correlation Analysis and aims to maximize the correlations between linear combinations of the model features. We show that our method of aligning models leads to better performances than past methods when averaging models trained on the same, or differing data splits. We also extend this analysis into the harder many models setting where more than 2 models are merged, and we find that CCA Merge works significantly better in this setting than past methods.
Accelerating programs is typically done by recognizing code idioms matching high-performance libraries or hardware interfaces. However, reco… (voir plus)gnizing such idioms automatically is challenging. The idiom recognition machinery is difficult to write and requires expert knowledge. In addition, slight variations in the input program might hide the idiom and defeat the recognizer. This paper advocates for the use of a minimalist functional array language supporting a small, but expressive, set of operators. The minimalist design leads to a tiny sets of rewrite rules, which encode the language semantics. Crucially, the same minimalist language is also used to encode idioms. This removes the need for hand-crafted analysis passes, or for having to learn a complex domain-specific language to define the idioms. Coupled with equality saturation, this approach is able to match the core functions from the BLAS and PyTorch libraries on a set of computational kernels. Compared to reference C kernel implementations, the approach produces a geometric mean speedup of 1.46× for C programs using BLAS, when generating such programs from the high-level minimalist language.
2024-03-02
2024 IEEE/ACM International Symposium on Code Generation and Optimization (CGO) (publié)
Artificial neural networks (ANNs) are considered "black boxes'' due to the difficulty of interpreting their learned weights.
While choosing… (voir plus) the best features is not well understood, random feature networks (RFNs) and wavelet scattering ground some ANN learning mechanisms in function space with tractable mathematics. Meanwhile, the genetic code has evolved over millions of years, shaping the brain to develop variable neural circuits with reliable structure that resemble RFNs. We explore a similar approach, embedding neuro-inspired, wavelet-like weights into multilayer RFNs. These can outperform scattering and have kernels that describe their function space at large width. We build learnable and deeper versions of these models where we can optimize separate spatial and channel covariances of the convolutional weight distributions. We find that these networks can perform comparatively with conventional ANNs while dramatically reducing the number of trainable parameters. Channel covariances are most influential, and both weight and activation alignment are needed for classification performance. Our work outlines how neuro-inspired configurations may lead to better performance in key cases and offers a potentially tractable reduced model for ANN learning.
Artificial neural networks (ANNs) are considered ``black boxes'' due to the difficulty of interpreting their learned weights.
While choosin… (voir plus)g the best features is not well understood, random feature networks (RFNs) and wavelet scattering ground some ANN learning mechanisms in function space with tractable mathematics. Meanwhile, the genetic code has evolved over millions of years, shaping the brain to devlop variable neural circuits with reliable structure that resemble RFNs. We explore a similar approach, embedding neuro-inspired, wavelet-like weights into multilayer RFNs. These can outperform scattering and have kernels that describe their function space at large width. We build learnable and deeper versions of these models where we can optimize separate spatial and channel covariances of the convolutional weight distributions. We find that these networks can perform comparatively with conventional ANNs while dramatically reducing the number of trainable parameters. Channel covariances are most influential, and both weight and activation alignment are needed for classification performance. Our work outlines how neuro-inspired configurations may lead to better performance in key cases and offers a potentially tractable reduced model for ANN learning.
