Benchmarking Project · Python

regescore-benchmark

benchmark your local llm

Master Mind@iboss21 · Atlas 72.5 · #2= builder

What it is

The regescore-benchmark repository implements a multi-stage gauntlet for benchmarking local LLMs, featuring procedurally generated tasks to avoid contamination and objective grading mechanisms to prevent self-grading bias. It provides both a command-line interface and a web dashboard for running evaluations and generating reports with confidence intervals.

Scored from the code at its last scan, 21 Oct 2026. How merit is scored

Where it counts

NLP & LLMs90%
Machine Learning40%
MLOps30%

How much of the project is about each field, as scoring counts it. It ranks in the field it is most about.

Around it on the NLP & LLMs board

Whole board

RankProjectBuilderFieldMerit

1ROCmForgeInference System · RustROCmForge is an LLM inference engine for AMD GPUs that implements custom HIP kernels for operations like attention, quantization, and RoPE, featuring a GPU Decode Graph to reduce kernel launch overhead and an inference path router for dynamic optimization.77.9/100ExpertLuiz Cesar SpiesNLP & LLMs
2regescore-benchmarkBenchmarking Project · Python76.5/100AdvancedMaster MindNLP & LLMs
3event-aware-SAE-trainerLibrary / Framework · PythonThe repository provides a library for training JumpReLU sparse autoencoders (SAEs) per transformer layer with an event-aware optimizer scheduler and augmented Lagrangian sparsity control. It includes tools for activation capture, checkpointing, and diagnostics, demonstrated by…75.2/100AdvancedRickNLP & LLMs
04RegesMoEML Application · C++RegesMoE is an end-to-end application that enables running the RegesCore 35B mixture-of-experts model on a normal PC with a single GPU, providing an installer, local OpenAI- and Anthropic-compatible API, and web app.72.5/100AdvancedMaster MindNLP & LLMs

What it is

Where it counts

Around it on the NLP & LLMs board

More from Master Mind