* add more tests to test_function * add function to llm * function decorator on llm * works * symbolic fixups * minimum change * implicit inputs * don't actually update llama yet
* start function and add walk rewrite * work * add function on feed_forward * llm progress * stuff * none of that