Themis Preference Pretrained Checkpoints Collection A collection of preference model pretraining checkpoints trained on general preference datasets intended as precursors for code reward models. • 6 items • Updated 2 days ago
Themis Preference Datasets & Benchmarks Collection A collection of preference datasets used for training and evaluation of code reward models. • 3 items • Updated 2 days ago
Themis Reward Model Collection Collection A collection of strong code reward models trained on a diverse collection of code preferences. • 6 items • Updated 2 days ago