Pinned Loading

1. Beyond-Early-Exit Beyond-Early-Exit Public

   Beyond Early Exit: Solving GPU Warp Divergence in Adaptive LLM Inference with Micro-Batched Routing. Author: Zachary Maronek Date: January 2026

   Python

2. fabq-rc fabq-rc Public

   FABQ-RC: Fisher-Adaptive Binary Quantization with Residual Codebooks

   Jupyter Notebook 3

3. particleblood-rust-gpu particleblood-rust-gpu Public

   Rust/WebGL2 GPU particle sim variant of particleblood

   Rust

4. ple-coded-gguf ple-coded-gguf Public

   PLE-Coded GGUF: exploiting Gemma E4B's per-layer embeddings as a compression side-channel

   Python

5. ingexuity ingexuity Public

   Julia 1

6. hgsel-moe hgsel-moe Public

   Hash-based Gradient-guided Sparse Expert Layer A deterministic, production-grade Sparse Mixture of Experts (MoE) architecture for dense Transformers.

   Python