Comment by dhorthy

3 months ago

Yeah I would hold that models don’t know how to simplify because most rl/benchmarks doesn’t penalize complexity

0 comments