On: Something weird is happening with LLMs and chess

[loa_in_]

Probably important when competing against undecimated ones from OpenAI

[NiloCK → loa_in_]

Notably: there were other OpenAI models that weren't quantize, but also performed poorly.