matryoshka-representation-learning

RAG Time: A 5-week Learning Journey to Mastering RAG

Official Code for Paper: Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation

Enterprise RAG ecosystem managing 32000+ semantic chunks. Features hybrid parsing (LlamaParse/PyMuPDF) and 256-dim MRL embeddings for 512MB RAM environments

Code and pretrained models for the paper: "MatMamba: A Matryoshka State Space Model"

Chem-MRL: SMILES-based Matryoshka Representation Learning Embedding Model

An investigation into how Matryoshka Representation Learning (MRL) with Relational Distillation affects the geometric structure and social bias encoding of Word2Vec embeddings.

Gradio app showcasing Chem-MRL embeddings for SMILES-based similarity search.

Matryoshka Representation Learning for CLIP image-text retrieval: nested InfoNCE makes embedding prefixes independently usable, cutting retrieval latency/storage. 64 dims keeps 95% of full-dim recall at 94% lower latency; cascade retrieval gives 22.4x speedup at >=99% recall.

Automated ICD coding system based on PLM-ICD, using Matryoshka Representation Learning (MRL) and Label Aware Attention to predict ICD codes based on text reports.

Application of Matryoshka Representation Learning on Text Embeddings

🚀 Build Ruby gems that utilize Rust for enhanced performance through two effective design patterns for seamless collaboration.