Shard is an AI assistant built with Rust, Tauri, and Typescript—supporting autonomous research tools, multimodal capabilities, and a 3‑tier personal memory system.

This paper introduces SynLexLM, a legal-domain LLM that scales via curriculum learning and synthetic data, improving legal QA and summarization while reducing fine-tuning and dataset demands.

This paper introduces LOTUS (LOttery Transformers with Ultra Sparsity), a novel method that leverages data lottery ticket selection and sparsity pruning to accelerate vision transformer training while maintaining accuracy.