hendel-gpt-j-6b-architecture

Summary

This records the internal dimensions of the GPT-J 6B model: how wide its internal representations are, how deep its processing stack is, and how many parallel "lenses" it uses to relate words to each other. It matters because these three numbers set the baseline for any fine-tuning, memory planning, or inference cost calculation the system might do with this model.