Benchmarks

• inference - a folder contains benchmark scripts that apply a throughput analysis for Llama models inference on various backends including on-prem, cloud and on-device.
• llm_eval_harness - a folder that introduces lm-evaluation-harness, a tool to evaluate Llama models including quantized models focusing on quality. We also included a recipe that calculates Llama 3.1 evaluation metrics Using lm-evaluation-harness and instructions that calculate HuggingFace Open LLM Leaderboard v2 metrics.