#pytorch

Matrix product scan with autograd and a hand-written backward
CodingHardOpenAI
▲ 00 solutions
Find the planted bugs in a small Transformer training script
CodingMediumOpenAI
▲ 00 solutions
Read and refactor a large PyTorch training file
CodingHardOpenAI
▲ 00 solutions
Debug a vision transformer notebook
CodingMediumApplied Intuition
▲ 00 solutions
Block-wise attention forward pass without the full score matrix
CodingHugging FacePerplexity
▲ 00 solutions
QKV attention programming with einsum fluency
ML designMediumAnthropic
▲ 01 solution