Sam Paech

slop-forensics. Jupyter Notebook

antislop-sampler. Python

auto-antislop. Python

diplobench. Benchmark for LLMs playing full press diplomacy

spiral-bench. HTML

antislop-vllm. Jupyter Notebook

slop-score. HTML

lm-evaluation-harness. A framework for few-shot evaluation of language models.

Ollama-MMLU-Pro-IRT. Ollama-MMLU-Pro fork, using a smaller IRT-tuned subset of MMLU-Pro

entropix-gsm8k-eval. Jupyter Notebook

not-x-but-y-bench. Python