arXiv Machine Learning By Aaryan Ajay Sharma, Sai Nishanth Padala, Seganrasan Subramanian

DARTS: Decoder-Aware Representation Tuning via Surgery for Model Merging

Read the original on arXiv Machine Learning →

The paper introduces DARTS, a method for tuning decoder representations during model merging. It addresses representation bias in autoregressive decoders by using an entropy‑weighted L1 loss and a per‑position additive bias to correct errors that accumulate across token positions. Experiments on code generation, mathematical reasoning, and instruction following with Llama‑2‑7B show that DARTS improves performance over standard surgery while adding only 0.1% extra parameters.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

arXiv AI
Jun 2

Video Reasoning without Training

arXiv:2510. 17045v2 Announce Type: replace-cross Abstract: Video reasoning using Large Multimodal Models (LMMs) relies on costly reinforcement learning (RL) and verbose chain-of-thought, resulting in substantial computational overhead during both training and inference.

By Deepak Sridhar, Kartikeya Bhardwaj, Jeya Pradha Jeyaraj, Nuno Vasconcelos, Ankita Nayak, Harris Teague
arXiv AI
Aug 28

Successive Capacity Growth: Task-Complexity-Driven Width and Depth Expansion for Vision Transformer Encoders in JEPA World Models

The paper introduces Successive Capacity Growth (SCG), a method for adaptively expanding Vision Transformer encoders in Joint-Embedding Predictive Architectures (JEPAs). SCG starts with a minimal encoder and incrementally increases width or depth based on a task‑agnostic test‑and‑verify mechanism, while a Sketched Isotropic Gaussian Regularizer (SIGReg) keeps learned semantic dimensions independent. Experiments on multi‑object dynamics and 2D navigation tasks show that SCG achieves up to 20.3% better prediction loss than fixed small baselines and 23% better than fixed large models, with far greater parameter efficiency and no false‑positive expansions.

By Frederik Berenz