About
Research & Development
Unmasking Hidden Biases: A New Method to Detect Spurious Correlations in AI Vision Models
TLDR: A new method called "Token Discarding" and metrics (A-TSI, M-TSI) are introduced to detect spurious correlations in Vision Transformers (ViTs). By removing image...
Research & Development
NER Retriever: A Flexible Framework for Finding Specific Entities in Text
TLDR: NER Retriever is a zero-shot framework for finding specific named entities in text based on user-defined type descriptions. It leverages internal, mid-layer representations...
Research & Development
The Hidden Vulnerability of LLMs: Performance Drops with Reworded Questions
TLDR: A new study reveals that while Large Language Model (LLM) rankings remain stable, their absolute performance significantly declines when benchmark questions are paraphrased....
Research & Development
Enhancing Video Compression with Smart Filtering Techniques
TLDR: This research introduces ConFRE, a novel approach to neural video compression that integrates two key filtering techniques: in-loop contextual filtering and out-of-loop reconstruction...
Research & Development
RepoDebug: A New Benchmark Reveals LLMs’ Struggles in Real-World Code Debugging
TLDR: RepoDebug is a novel, multi-task, and multi-language dataset designed to evaluate Large Language Models (LLMs) on complex, repository-level code debugging. It features 22...
Research & Development
Speech Models Uncover Secrets in Animal Sounds: A New Approach to Bioacoustics
TLDR: This research paper demonstrates that self-supervised speech models (like HuBERT, WavLM, XEUS) can effectively transfer their learned acoustic representations to various animal bioacoustic...