Update ppl evals to work with blt model, in addition to entropy model (#82) 083656c unverified par-meta commited on Mar 13
Let process start before yielding preloaded prefetch buffer, avoid needlessly losing buffer in edge cases (#75) 8f2cf88 unverified par-meta commited on Mar 5
Correctly reset batch iterator at each arrow create_iter call. (#74) c727844 unverified par-meta commited on Mar 4
Pass mask in packing_iterator, correctly handle last batch, fix masking (#65) 08b8c7c unverified par-meta commited on Feb 27
Remove byte tokenizer and add config args to switch between byte/patch packing (#68) aeb95f1 unverified par-meta commited on Feb 25
Update iterator inheritance, pass file format args, limit iterator (#63) fc3399e unverified par-meta commited on Feb 22
Fix multiprocessing dataloader checkpointing and use it in the train script (#50) 8c61ab5 unverified par-meta commited on Feb 13
This includes fixes that make checkpointing and reloading work correctly. (#35) 7044771 unverified par-meta commited on Jan 28
Initial codes and scripts for training entropy model (#34) 7622d28 unverified par-meta commited on Jan 27
Changes for training entropy model and correcting attention in local models (#25) 6ffeb66 unverified par-meta commited on Jan 17
Replace regular filesystem calls with fsspec + add s3 support (#18) b0120da unverified par-meta commited on Jan 10