Code

History

Commits on move/encoder_kernel35+
  1. also move the launchersAndrej Karpathy committed 2 years ago49b9776
  2. move encoder, kernel 1/NAndrej Karpathy committed 2 years ago93546ae
  3. Merge pull request #535 from ngc92/speed-up-compileAndrej committed 2 years ago3392671
  4. moved device utilities into a separate file, and follow-up cleanupErik Schultheis committed 2 years ago67502c7
  5. Merge pull request #528 from gordicaleksa/refactor_evalloaderAndrej committed 2 years ago8cc39a7
  6. Merge pull request #529 from gordicaleksa/improve_start_training_print_infoAndrej committed 2 years ago2a6a5b2
  7. Merge pull request #530 from gordicaleksa/minor_refactorAndrej committed 2 years ago0cd9f1a
  8. Merge pull request #532 from ngc92/ln-buffersAndrej committed 2 years agod17893b
  9. Merge pull request #533 from karpathy/feature/refactor_part2Andrej committed 2 years agoac93145
  10. resolve deviceProp and cudaCheck dupesAndrej Karpathy committed 2 years ago001ad78
  11. small more fixesAndrej Karpathy committed 2 years ago020b15a
  12. some fixes, still not doneAndrej Karpathy committed 2 years ago5d94034
  13. refactor part 2, moving stuff into common files so that we can later nicely separate out all the kernels as wellAndrej Karpathy committed 2 years ago804d069
  14. Merge pull request #526 from karpathy/try/move_cudnn_to_llmcAndrej committed 2 years agoe402684
  15. more efficient use of memory buffers for LN recomputationErik Schultheis committed 2 years ago0d16b51
  16. attempt to remove PATH_SEP from MakefileAndrej Karpathy committed 2 years ago6434009
  17. Minor refactor: remove hardcoded val, delete unused vars...Aleksa Gordic committed 2 years ago42ca620
  18. Remove unnecessary load filename lineAleksa Gordic committed 2 years ago7413a1a
  19. Improve start of training print infoAleksa Gordic committed 2 years ago728762c
  20. Remove file size - not usedAleksa Gordic committed 2 years agod9e8676
  21. Refactor eval loader - remove 2 unnecessary malloc calls per iterationAleksa Gordic committed 2 years agoe86bd62
  22. attempt to move cudnn attention to the llmc directoryAndrej Karpathy committed 2 years ago41dc65a
  23. delete copy paste code for gpt2 model initAndrej Karpathy committed 2 years ago08fc45b
  24. Merge branch 'ngc92-mfu'Andrej Karpathy committed 2 years agoc827dd2
  25. add new file llmc/mfu.h for MFU calculations for different GPUsAndrej Karpathy committed 2 years ago1ac3059
  26. Merge pull request #512 from gordicaleksa/refactor_encoder_bwd_kernelAndrej committed 2 years ago80e0c3b
  27. Merge pull request #514 from ngc92/compile-fixAndrej committed 2 years agocb5eff0
  28. Merge pull request #515 from gordicaleksa/fix_adamw_kernel3_num_threadsAndrej committed 2 years agoff98d8b
  29. Merge pull request #516 from gordicaleksa/update_random_docAndrej committed 2 years ago1ce30fd
  30. Merge pull request #519 from gordicaleksa/mem_leakAndrej committed 2 years ago2525849
  31. Merge branch 'gordicaleksa-fix_zero_stage1'Andrej Karpathy committed 2 years ago1e893f4
  32. adjust commentsAndrej Karpathy committed 2 years ago4944ac5
  33. Add ; - ah c my old friend :)Aleksa Gordic committed 2 years agoa8a5a44
  34. Improve commentsAleksa Gordic committed 2 years ago9a455ed
  35. Fix zero stage 1 norm computation - WIPAleksa Gordic committed 2 years agob4e8bd7