分支列表

master

f1e2ace651 · Merge pull request #801 from ngc92/ngc92/fix-test · 最后更新于 2025-05-10 23:24:10 +00:00

llama3 da88cb11b0 · Merge pull request #803 from ngc92/ngc92/llama3-tied-weights · 最后更新于 2025-06-26 17:03:39 +00:00 frostbyte_neo	6 75		ZIP TAR.GZ
feature/managed2 18298f3a40 · i misspelled reduced · 最后更新于 2024-08-16 18:16:46 +00:00 frostbyte_neo	21 0	已包含	ZIP TAR.GZ
gordicaleksa-fix_dataloader2 755458d0e1 · fix tokenizer omg · 最后更新于 2024-08-13 17:24:15 +00:00 frostbyte_neo	29 0	已包含	ZIP TAR.GZ
feature/finetune_llama31py 0aa9175569 · still wip just putting things up for comment · 最后更新于 2024-08-04 20:26:41 +00:00 frostbyte_neo	82 1		ZIP TAR.GZ
feature/restore_from_master 51dd102328 · remove confusing comment · 最后更新于 2024-07-30 20:01:43 +00:00 frostbyte_neo	97 0	已包含	ZIP TAR.GZ
refactor/code-to-zerocuh 5afc57064c · document the imports from zero.cuh to be explicit · 最后更新于 2024-07-15 19:09:52 +00:00 frostbyte_neo	133 0	已包含	ZIP TAR.GZ
feature/gpt3v1 220125f39c · ok let's allow T < max_seq_len because our tests use it. make warning very prominent · 最后更新于 2024-07-15 18:21:34 +00:00 frostbyte_neo	139 0	已包含	ZIP TAR.GZ
fix/reshuffle_zero_grad 7b92930021 · move things around a bit, a few comments, and also update python script to correspond to C more · 最后更新于 2024-07-01 17:20:46 +00:00 frostbyte_neo	180 0	已包含	ZIP TAR.GZ
overfit-batch-actually 7fd7717816 · fix the overfit single batch behavior to actually overfit batch, not microbatch, we do this more cleanly by simply resetting the dataloader every step · 最后更新于 2024-07-01 17:00:40 +00:00 frostbyte_neo	187 0	已包含	ZIP TAR.GZ
feature/openmpi_flexible_path 7c283df9c4 · use ?= for OPENMPI_DIR so it can be overriden · 最后更新于 2024-06-26 16:06:54 +00:00 frostbyte_neo	213 0	已包含	ZIP TAR.GZ
feature/outliers e12d090483 · resolve merge conflicts · 最后更新于 2024-06-26 04:05:46 +00:00 frostbyte_neo	216 0	已包含	ZIP TAR.GZ
feature/rolling_checkpoints 11e2fd1cf2 · separate out write_checkpoint and delete_checkpoint · 最后更新于 2024-06-25 20:46:03 +00:00 frostbyte_neo	250 0	已包含	ZIP TAR.GZ
feature/lr_schedulers dbe06082ca · delete spurious check its fine · 最后更新于 2024-06-21 02:39:40 +00:00 frostbyte_neo	292 1		ZIP TAR.GZ
feature/nccl-only f19202c858 · small fixes on top of @chinthysl PR to kick off the feature branch · 最后更新于 2024-06-20 20:13:27 +00:00 frostbyte_neo	307 3		ZIP TAR.GZ
feature/hf_eval 1a66151a80 · small fixes and readme · 最后更新于 2024-06-19 01:03:36 +00:00 frostbyte_neo	403 0	已包含	ZIP TAR.GZ
feature/permuted_dataloader 14d9cd671e · resolve merge conflict in staate file between use_master_weights and DataLoader should_shuffle · 最后更新于 2024-06-17 19:16:42 +00:00 frostbyte_neo	354 0	已包含	ZIP TAR.GZ
feature/delete_backward_act cbe82e783a · consolidate memory by deleting the grad activation struct, move some of the last pieces of memory to the forward pass · 最后更新于 2024-06-14 00:54:58 +00:00 frostbyte_neo	423 0	已包含	ZIP TAR.GZ
feature/overlap-backward-reduce 43a903e1b4 · revert old printf0, minor touchups and comments, add assert · 最后更新于 2024-06-13 23:18:22 +00:00 frostbyte_neo	425 0	已包含	ZIP TAR.GZ
feature/streams ee6b3c9325 · take out Async copies and memsets · 最后更新于 2024-06-08 16:17:02 +00:00 frostbyte_neo	479 0	已包含	ZIP TAR.GZ
move/matmul 094988738b · adjust the comment on extern deviceProp to be more descriptive · 最后更新于 2024-06-03 22:57:00 +00:00 frostbyte_neo	506 0	已包含	ZIP TAR.GZ

1 2 3

默认分支

分支列表