adzcai comments on Video/​animation: Neel Nanda explains what mechanistic interpretability is

[adzcai]

Or even better, finetuning an LLM to automate writing the code!

[the gears to ascension]

cyborgism, activate!

just don’t use an overly large model.