Show HN: GPT-2 inference in pure C#, 0 bytes allocated per token(github.com)
github.com
Show HN: GPT-2 inference in pure C#, 0 bytes allocated per token
https://github.com/DevOnBike/Overfit
2 コメント
For me, the interesting part here isn't GPT-2, it's the memory discipline. I feel like most inference runtimes slowly leak allocations everywhere as features pile up.