Redditor suggests post-training over benchmarking for local LLM hardware

View on Reddit

A Reddit post argues that users with local hardware should explore post-training instead of simply benchmarking model speed. The author calls benchmarking "boring and intellectually lazy" and urges deeper engagement with LLM customization.