This is the starter code for CSE493S HW 1 for Spring 2026. For a complete description of the homework please see ps.md or ps.pdf. Please contact the TAs if you find any bugs in this repo.
The file model.py contains an implementation of a transformer model. The file part_0_1_contract.py contains some function signatures that would make autograding less painful for the TAs.
The notebook part_2_starter.ipynb has code to load pretrained models, the AIME dataset, functions for evaluation and has unoptimized code for inference. We encourage you to write your own inference code to speed things up.