Ctrl+K
- 2024-06-06_AdamW
- 2024-10-09_SOAP
- 2024-10-10_Muon
- 2024-10-13_llmc
- 2024-10-14_ModernArch
- 2024-10-17_DistributedMuon
- 2024-10-18_PyTorch25
- 2024-10-20_ScaleUp1B
- 2024-10-29_Optimizers
- 2024-11-03_UntieEmbed
- 2024-11-04_50Bruns
- 2024-11-06_ShortcutsTweaks
- 2024-11-08_CastBf16
- 2024-11-09_Replicateleloykun
- 2024-11-10_ScaleShortcuts
- 2024-11-10_UNetDoubleLr
- 2024-11-14_QuantizedFP4
- 2024-11-19_FlexAttention
- 2024-11-24_WindowWarmup
- 2024-12-04_ValueEmbed
- 2024-12-08_UNetValueEmbedsTweaks
- 2024-12-10_MFUTweaks
- 2024-12-17_SparsifyEmbeds
- 2025-01-04_SoftCap
- 2025-01-13_Fp8LmHead
- 2025-01-16_Sub3Min
- 2025-01-26_BatchSize
- 2025-02-01_RuleTweak
- 2025-05-09_SkipMLPBlocks
- 2025-05-24_FasterReduce
- 2025-05-24_StableTorch
- 2025-05-25_EvenFasterReduce
- 2025-05-25_MuonWithAuxAdamExample
- 2025-05-30_noallreduce
- 2025-07-12_BosAlign
- 2025-07-13_UpgradeTorch190
- 2025-07-18_TritonMuon
- 2025-08-23_SparseAttnGate
- 2025-09-03_FA3
- 2025-09-05_SkipMLPBlocks
- 2025-09-10_Yarn
- 2025-09-11_VectSigmoidBFloat16
- 2025-09-15_AsyncDataLoadAttnFinalWindow
- 2025-09-18_Smear
- 2025-09-21_DropAttn
- 2025-09-23_MuonCustomSizing
- 2025-09-27_BF16CE
- 2025-09-29_PolarExpress
- 2025-09-30_CustomBatching
- 2025-10-04_Backout