Code

History

Commits on master1536
  1. Merge pull request #801 from ngc92/ngc92/fix-testAndrej committed 1 year agof1e2ace
  2. adjusted error thresholdsErik Schultheis committed 1 year ago42aa891
  3. reorder checks to match actual tensor orderErik Schultheis committed 1 year ago166f5e6
  4. Merge pull request #766 from msaroufim/patch-1Andrej committed 2 years ago7ecd890
  5. Update README.mdMark Saroufim committed 2 years ago315b8d1
  6. cuda mode -> gpu modeMark Saroufim committed 2 years ago5b2e318
  7. change default params: use tinyshakespeare and decrease LRAndrej Karpathy committed 2 years agobd8c604
  8. Merge pull request #724 from GaoYusong/masterAndrej committed 2 years agobd457aa
  9. Merge pull request #733 from zhangpiu/feature/llm.cppAndrej committed 2 years agoa2bdae2
  10. Merge pull request #735 from gordicaleksa/minor_refactor3Andrej committed 2 years agoebc28b9
  11. Merge pull request #744 from dengl11/prAndrej committed 2 years ago2c92137
  12. Merge pull request #745 from karpathy/feature/managed2Andrej committed 2 years ago0ddedf9
  13. i misspelled reducedAndrej Karpathy committed 2 years ago18298f3
  14. reduce across GPUs nicerAndrej Karpathy committed 2 years ago8c586f9
  15. fallback to memory allocation of m,v,master_weights on host automatically in case of OOM. will run slower but won't OOMAndrej Karpathy committed 2 years agoe6856bc
  16. fix makefile for multigpu setupsAndrej Karpathy committed 2 years ago2882ec6
  17. hint to hostErik Schultheis committed 2 years agof72c1f2
  18. just try to allocate on device; fallback if that failsErik Schultheis committed 2 years agoc845757
  19. fall back to cudaMallocManaged for optimizer states if we're out of memoryErik Schultheis committed 2 years ago0d52d2a
  20. fix a typoLi Deng committed 2 years agoff356e4
  21. Merge pull request #740 from karpathy/gordicaleksa-fix_dataloader2Andrej committed 2 years ago4c84bc7
  22. fix tokenizer omgAndrej Karpathy committed 2 years ago755458d
  23. attempt to fix PRAndrej Karpathy committed 2 years ago16635d4
  24. Merge branch 'fix_dataloader2' of https://github.com/gordicaleksa/llm.c into gordicaleksa-fix_dataloader2Andrej Karpathy committed 2 years ago9740a65
  25. Merge pull request #738 from ademeure/faster_compileAndrej committed 2 years ago1787210
  26. Refactor tokenize, delegate insteadAleksa Gordic committed 2 years ago68624cc
  27. Refactor magic/version header dictAleksa Gordic committed 2 years agoe8739c8
  28. Improve compile time via nvidia-smi instead of __nvcc_device_query (& add "--threads=0").ademeure committed 2 years agofc88344
  29. Improve compile time via nvidia-smi instead of __nvcc_device_query (& add "--threads=0").ademeure committed 2 years ago9b1c14b
  30. Replace fire with argparseAleksa Gordic committed 2 years agoeb1b359
  31. Add fire to reqsAleksa Gordic committed 2 years ago311c149
  32. Add fineweb llama 3 supportAleksa Gordic committed 2 years ago5076563
  33. Use fire module, update readmeAleksa Gordic committed 2 years agof8c1ff5
  34. Add tiny shakespeareAleksa Gordic committed 2 years ago15b76bf
  35. Change default dataset path to tiny storiesAleksa Gordic committed 2 years agob018b0c