Image de couverture de livre

Task Priors: Enhancing Model Evaluation by Considering the Entire Space of Downstream Tasks

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Patel, Niket, Balestriero, Randall
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

SAFE: A Novel Approach to AI Weather Evaluation through Stratified Assessments of Forecasts over Earth
par: Masi, Nick, et autres
Publié: (2025)

No Location Left Behind: Measuring and Improving the Fairness of Implicit Representations for Earth Data
par: Cai, Daniel, et autres
Publié: (2025)

ALLoRA: Adaptive Learning Rate Mitigates LoRA Fatal Flaws
par: Huang, Hai, et autres
Publié: (2024)

Why and How Auxiliary Tasks Improve JEPA Representations
par: Yu, Jiacan, et autres
Publié: (2025)

Fast and Exact Enumeration of Deep Networks Partitions Regions
par: Balestriero, Randall, et autres
Publié: (2024)

The Fair Language Model Paradox
par: Pinto, Andrea, et autres
Publié: (2024)

Occam's Razor for Self Supervised Learning: What is Sufficient to Learn Good Representations?
par: Ibrahim, Mark, et autres
Publié: (2024)

Characterizing Large Language Model Geometry Helps Solve Toxicity Detection and Generation
par: Balestriero, Randall, et autres
Publié: (2023)

SlotFM: A Motion Foundation Model with Slot Attention for Diverse Downstream Tasks
par: Park, Junyong, et autres
Publié: (2025)

Downstream Task-Oriented Generative Model Selections on Synthetic Data Training for Fraud Detection Models
par: Cheng, Yinan, et autres
Publié: (2024)

GPS-SSL: Guided Positive Sampling to Inject Prior Into Self-Supervised Learning
par: Feizi, Aarash, et autres
Publié: (2024)

FedEval-LLM: Federated Evaluation of Large Language Models on Downstream Tasks with Collective Wisdom
par: He, Yuanqin, et autres
Publié: (2024)

Position: An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning Research
par: Reizinger, Patrik, et autres
Publié: (2025)

WxC-Bench: A Novel Dataset for Weather and Climate Downstream Tasks
par: Shinde, Rajat, et autres
Publié: (2024)

Modular Deep Learning for Multivariate Time-Series: Decoupling Imputation and Downstream Tasks
par: Raj, Joseph Arul, et autres
Publié: (2024)

LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics
par: Balestriero, Randall, et autres
Publié: (2025)

Learning by Reconstruction Produces Uninformative Features For Perception
par: Balestriero, Randall, et autres
Publié: (2024)

VocabTailor: Dynamic Vocabulary Selection for Downstream Tasks in Small Language Models
par: Zhang, Hanling, et autres
Publié: (2025)

Entire Space Counterfactual Learning for Reliable Content Recommendations
par: Wang, Hao, et autres
Publié: (2022)

Task-Informed Anti-Curriculum by Masking Improves Downstream Performance on Text
par: Jarca, Andrei, et autres
Publié: (2025)

LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels
par: Maes, Lucas, et autres
Publié: (2026)

Stable Diffusion Dataset Generation for Downstream Classification Tasks
par: Lomurno, Eugenio, et autres
Publié: (2024)

Self-Supervised Anomaly Detection in the Wild: Favor Joint Embeddings Methods
par: Otero, Daniel, et autres
Publié: (2024)

PrAg-PO: Prompt Augmented Policy Optimization for Robust and Diverse Mathematical Reasoning
par: Lu, Wenquan, et autres
Publié: (2026)

Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks
par: Li, Miaomiao, et autres
Publié: (2025)

Enabling Agents to Communicate Entirely in Latent Space
par: Du, Zhuoyun, et autres
Publié: (2025)

Understanding and Mitigating the Label Noise in Pre-training on Downstream Tasks
par: Chen, Hao, et autres
Publié: (2023)

Learning to Rewrite Prompts for Bootstrapping LLMs on Downstream Tasks
par: Zhou, Qinhao, et autres
Publié: (2025)

Shaping the Prior: How Synthetic Task Distributions Determine Tabular Foundation Model Quality
par: Bouadi, Mohamed, et autres
Publié: (2026)

Junk DNA Hypothesis: Pruning Small Pre-Trained Weights Irreversibly and Monotonically Impairs "Difficult" Downstream Tasks in LLMs
par: Yin, Lu, et autres
Publié: (2023)

Joint Embedding vs Reconstruction: Provable Benefits of Latent Space Prediction for Self Supervised Learning
par: Van Assel, Hugues, et autres
Publié: (2025)

Deep Networks Always Grok and Here is Why
par: Humayun, Ahmed Imtiaz, et autres
Publié: (2024)

On the Geometry of Deep Learning
par: Balestriero, Randall, et autres
Publié: (2024)

LoRA Users Beware: A Few Spurious Tokens Can Manipulate Your Finetuned Model
par: Salles, Marcel Mateos, et autres
Publié: (2025)

DIVERGE: Diversity-Enhanced RAG for Open-Ended Information Seeking
par: Hu, Tianyi, et autres
Publié: (2026)

Entire Chain Uplift Modeling with Context-Enhanced Learning for Intelligent Marketing
par: Huang, Yinqiu, et autres
Publié: (2024)

Task-Aware Virtual Training: Enhancing Generalization in Meta-Reinforcement Learning for Out-of-Distribution Tasks
par: Kim, Jeongmo, et autres
Publié: (2025)

Efficient Embedding-based Synthetic Data Generation for Complex Reasoning Tasks
par: Jayaraman, Srideepika, et autres
Publié: (2026)

Reconciling Model Multiplicity for Downstream Decision Making
par: Du, Ally Yalei, et autres
Publié: (2024)

Gaussian Embeddings: How JEPAs Secretly Learn Your Data Density
par: Balestriero, Randall, et autres
Publié: (2025)