Why acc and acc_norm disagree in lm-eval-harness: log-likelihood length bias, byte-length normalization, and when each multiple-choice metric lies.
acc vs acc_norm: Why Length Bias Skews LLM Eval Scores
Related
I built an AI Shopify audit tool, then decided to take it back into my own hands
Most Shopify audit tools give you a polite report. Vetta roasts your store instead and tells you...
Speculative Decoding and MTP: Why Guessing Is Free
A forward pass over five tokens costs about the same as over one. That gap is the entire speedup, and MTP is how the model drafts guesses to fill it.
How to Create Your Own Claude Code Skill With SKILL.md
If you use Claude Code for frontend development, you may have noticed something. Claude can write...