paper-with-me

홈 › Papers

Representation Without Reward: A JEPA Audit for LLM Fine-Tuning

2026-05-14 · Biswa Sengupta arxiv

Joint-embedding predictive architectures (JEPAs) propose that a model should learn more useful abstractions when trained to predict latent representations rather than observed outputs. For autoregressive language-model fine-tuning the principle entails a stricter requirement: the induced hidden-state geometry must reach the language-model head \emph{and} improve the decoded task metric. We test that requirement under a fixed Llama-3.2-1B-Instruct LoRA harness on natural-language-to-regex generation, comparing twenty-two training-time auxiliaries across trajectory-shape regularisation, distributional constraints, predictor/target asymmetry, Fisher-metric Jacobi residuals, and a decoder-visible JEPA objective constructed to lie in cross-entropy's positive cone. The empirical answer is a structured null: several auxiliaries clear single-cell paired $α= 0.10$ without correction (T3-Local at $Δ= +2.53$~pp, $p = 0.003$ being the strongest), but none survives Bonferroni or Holm--Bonferroni at the relevant family-wise threshold, even though many change curvature, anisotropy, variance, and gradient direction. Decoder-visible JEPA yields the first positive auxiliary--cross-entropy gradient cosine in the study, yet exact match remains inside seed noise; a full-fine-tuning replication of the same auxiliary at $n = 5$ seeds reproduces the null on both benchmarks (TURK: $Δ= +0.04$~pp, $p_{\text{paired}} = 0.96$; SYNTH: $Δ= +0.52$~pp, $p_{\text{paired}} = 0.28$), so the null is robust across LoRA and full fine-tuning for the decoder-visible construction. Hidden-state representation work and decoded-task accuracy are therefore weakly coupled in this regime; we accordingly reframe LLM-domain JEPA evaluation as a coupling problem, in which the operative question is under which metrics useful hidden geometry becomes decoder-visible task signal.

📄 PDF Abstract BibTeX arXiv:2605.15394

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

JEPA Policy: Diffusion-Free Imitation Learning via Paired Action and Future Representation Prediction

2026-09-09 · Jie Xu, Kangjin Yu, Ziyi Jin, Junjie Gao 외 arxiv

Standard behavior cloning supervises actions without explicitly constraining the future representation paired with each demonstrated action chunk. We introduce JEPA Policy, a diffusion-free framework that uses the action…

TD-JEPA: Latent-predictive Representations for Zero-Shot Reinforcement Learning

2025-10-01 · Marco Bagatella, Matteo Pirotta, Ahmed Touati, Alessandro Lazaric 외 arxiv

Latent prediction--where agents learn by predicting their own latents--has emerged as a powerful paradigm for training general representations in machine learning. In reinforcement learning (RL), this approach has been e…

Reinforcement Learning

Joint Embedding Predictive Architectures Focus on Slow Features

2022-11-20 · Vlad Sobal, Jyothir S V, Siddhartha Jalagam, Nicolas Carion 외

Many common methods for learning a world model for pixel-based environments use generative architectures trained with pixel-level reconstruction objectives. Recently proposed Joint Embedding Predictive Architectures (JEP…

Predictive Objectives Discard Exogenous Control-Relevant Features: A Controlled Mechanistic Study

2026-06-29 · Ayan Pendharkar arxiv

Joint-embedding predictive (JEPA-style) objectives learn representations by predicting future latents. In doing so they can discard features that are exogenous (uncontrollable by the agent) yet control-relevant, even whe…

WA-JEPA: Rethinking the Video JEPA Paradigm for World-Action Modeling in Autonomous Driving

2026-08-21 · Xinlin Wang, Yujiao Xiang, Yuheng Zhou, Jingqi Wang 외 arxiv

Video Joint Embedding Predictive Architecture (V-JEPA) learns powerful spatiotemporal representations from video through self-supervised latent feature prediction. However, V-JEPA is built around random-mask completion a…

Autonomous Driving