Semantic Compositions Enhance Vision-Language Contrastive Learningby Maxwell Aladago, Lorenzo Torresani, Soroush VosoughiFirst submitted to arxiv…
FastCLIP: A Suite of Optimization Techniques to Accelerate CLIP Training with Limited Resourcesby Xiyuan Wei,…
Can Small Language Models Learn, Unlearn, and Retain Noise Patterns?by Nicy Scaria, Silvester John Joseph…
Fine-tuned network relies on generic representation to solve unseen cognitive taskby Dongyan LinFirst submitted to…
Grass: Compute Efficient Low-Memory LLM Training with Structured Sparse Gradientsby Aashiq Muhamed, Oscar Li, David…
Data curation via joint example selection further accelerates multimodal learningby Talfan Evans, Nikhil Parthasarathy, Hamza…
BlockLLM: Memory-Efficient Adaptation of LLMs by Selecting and Optimizing the Right Coordinate Blocksby Amrutha Varshini…
Segment Any Text: A Universal Approach for Robust, Efficient and Adaptable Sentence Segmentationby Markus Frohmann,…
ReCaLL: Membership Inference via Relative Conditional Log-Likelihoodsby Roy Xie, Junlin Wang, Ruomin Huang, Minxing Zhang,…
Uni-Mol2: Exploring Molecular Pretraining Model at Scaleby Xiaohong Ji, Zhen Wang, Zhifeng Gao, Hang Zheng,…