programmer

Joined August 2021
Jay Mody retweeted
This is still the best* introduction to GPTs I have ever read. jaykmody.com/blog/gpt-from-s… Fills in all the gaps between Attention is All You Need right up to coding every single line of a working GPT2 implementation yourself. Even ends with explanations of finetuning for classification, summarization, instruction tuning, and PEFT. Bravo @jaykmody and all who take the time to write these things down in fully commented python rather than cryptic latex. *best WRITTEN introduction. we all know who has the best video intro to GPTs…
16
205
5
1,528
222,785
Late to the party but "GPT in 60 Lines of NumPy" / picoGPT is nicely done: jaykmody.com/blog/gpt-from-s… - good supporting links/pointers - flexes some of the benefits of JAX: 1) trivial to port numpy -> jax.numpy, 2) get gradients, 3) batch with jax.vmap - inferences gpt-2 checkpoints
15
134
5
1,348
222,598