KV-cache compression for LLMs: reference implementations of TurboAngle and TurboQuant codecs with Triton GPU kernels

Open source implementation and extension of Google Research’s PaperBanana for automated academic figures, diagrams, and research visuals, expanded to new domains like slide generation.

ParamΔ (Parameter Delta) Implementation - Enabling Continuous Model Post-Training at No Training Cost

Fast LLM inference with 2.8x speedup using speculative decoding

Implementation and evaluation of Scaling Embedding Layers in Language Models research paper

Implementation of Better Embeddings with Coupled Adam research paper

Implementation of research paper: DarwinLM: Evolutionary Structured Pruning of Large Language Models

Implementing ReaRAG, a knowledge-guided reasoning model that enhances factual accuracy using iterative retrieval-augmented generation. Adapting the methodology for modular LLM integration and customizable retrieval strategies.

Anytime Verified Agents (AVA) is a framework for building reliable LLM-based reasoning systems that adaptively allocate computational resources under budget constraints. AVA combines adaptive search strategies, uncertainty estimation, and verification cascades to maximize reliability while respecting token, tool-call, and verification budgets.

Power BI to Snowflake SQL translator with enhanced M and DAX support