Commit History

Autor SHA1 Mensaxe Data
  Tri Dao bedf877467 [CrossEntropy] Fix where labels address not aligned to 16 bytes hai 2 meses
  Tri Dao 8c20cfef49 [Rotary] Support qkv block layout from GQA hai 3 meses
  Tri Dao c7f32a8409 [CrossEntropy] Support precomputed LSE hai 3 meses
  Tri Dao d79f9b41a8 [CrossEntropy] Use online softmax to simplify implementation hai 3 meses
  lancerts 22339db185 remove an unused import (#960) hai 6 meses
  Tri Dao ec6d22143b [CrossEntropy] Change ignored_index -> ignore_index hai 7 meses
  Curtis "Fjord" Hawthorne d8aacc510c return z_loss (#768) hai 10 meses
  Tri Dao 08124c8f9c [CrossEntropy] Implement logit_scale option hai 1 ano
  Tri Dao aaa1474129 [CrossEntropy] Simplify the case of large vocab with Tensor Parallel hai 1 ano
  Shijie abf04a56e1 fix flash ce mp large vocab (#673) hai 1 ano
  Tri Dao c79de85ffa [CrossEntropy] Fix triton cross_entropy_loss IMA for >=2B elements hai 1 ano
  Tri Dao 5400fdc4ac [CE] Implement CrossEntropyLoss in Triton hai 1 ano