Research papers,
explained for humans.

Sensitivity-preserving of Fisher Information Matrix through random data down-sampling for experimental design

A Metamorphic Testing Perspective on Knowledge Distillation for Language Models of Code: Does the Student Deeply Mimic the Teacher?

This paper introduces MetaCompress, a metamorphic testing framework that reveals significant behavioral discrepancies between teacher and student code language models under adversarial conditions—discrepancies missed by traditional accuracy metrics—and demonstrates its effectiveness in evaluating the behavioral fidelity of models compressed via knowledge distillation.

Multimodal Diffusion Forcing for Forceful Manipulation

This paper introduces Multimodal Diffusion Forcing (MDF), a unified framework that leverages random partial masking and diffusion models to learn rich temporal and cross-modal dependencies from expert trajectories, thereby achieving robust and versatile performance in contact-rich, forceful manipulation tasks.

Locally Trivial Deformations of Toric Varieties

This paper introduces a combinatorial deformation functor $\mathrm{Def}_\Sigma$ based on Čech cochains that is isomorphic to the locally trivial deformation functor of toric varieties under specific conditions, enabling new criteria for unobstructedness, explicit obstruction formulas, and a classification of unobstructed iterated $\mathbb{P}^1$-bundle threefolds.

Enumeration of dihypergraphs with specified degrees and edge types

This paper provides asymptotic formulae for enumerating directed hypergraphs with specified in-degree and out-degree sequences, as well as fixed head and tail sizes for all hyperarcs, under the condition that the maximum values of these parameters remain sufficiently small.

SVD-Prune: Training-Free Token Pruning For Efficient Vision-Language Models

SVD-Prune is a training-free, plug-and-play token pruning method that utilizes Singular Value Decomposition and statistical leverage scores to select the most informative vision tokens, effectively overcoming the limitations of existing heuristic-based approaches to maintain high performance in Vision-Language Models even under extreme token budget constraints.

RL makes MLLMs see better than SFT

This paper demonstrates that Reinforcement Learning (RL) significantly outperforms Supervised Fine-Tuning (SFT) in enhancing Multimodal Large Language Models by fundamentally reshaping their vision encoders to produce stronger, more localized visual representations, leading to the proposal of a computationally efficient training framework called Preference-Instructed Vision OpTimization (PIVOT).

From Translation to Superset: Benchmark-Driven Evolution of a Production AI Agent from Rust to Python

This paper presents a benchmark-driven methodology for migrating a 648K-line production Rust AI agent to Python, demonstrating that LLM-assisted translation guided by public benchmarks not only achieves functional parity but also enables the creation of a more expressive, feature-rich superset with significantly reduced code volume.