Contents Menu Expand Light mode Dark mode Auto light/dark, in light mode Auto light/dark, in dark mode Skip to content
ExactRef
ExactRef
  • Get started
  • Guide
    • How tolerance works
    • Reading complicated tolerances
    • Results
    • Data in and out
    • Backward
    • Chains and any_of
    • Algorithms
    • Guarantees and limits
  • Solutions
    • Replace hand-picked test tolerances
    • Test a custom kernel
    • Check a PyTorch operation across dtypes
    • Check low-precision kernels
    • Accept nondeterministic reductions
    • Check gradients and custom backward kernels
    • Check fused attention kernels
    • Compare the accuracy of two implementations
    • Find the step that loses accuracy
    • Decide whether inf or NaN is legitimate
    • Generate golden reference values
    • Learn and teach floating-point rounding
  • Reference
    • Operations
      • matmul
      • gemm
      • matrix_exp and matrix_exp_frechet
      • add and multiply
      • sum
      • scaled dot-product attention
    • Results
      • Result
      • ExactArray and Enclosure
      • any_of
    • Data
      • Arrays and values
      • Formats and rounding
    • Algorithms
    • PyTorch equivalents
    • Errors
  • Glossary
  • Contact
  • GitHub
  • PyPI
Made possible by My Financing USA
Back to top
MIT License GitHub PyPI Contact