Skip to content

GitLab

Switch branch/tag
  1. 12 Jul, 2019 1 commit
  3. 03 Jul, 2019 2 commits
  5. 28 Jun, 2019 1 commit
  7. 14 Jun, 2019 1 commit
  9. 11 Jun, 2019 1 commit
  11. 31 May, 2019 2 commits
  13. 26 Apr, 2019 1 commit
    • ptrblck's avatar
      Replace type().ScalarType() with scalar_type() (#272) · 855808f3
      ptrblck authored 
      * change .type().ScalarType() to .scalar_type() + at::ScalarType::X to at::kX

* revert scalar_type() to type() for AT_DISPATCH_FLOATING_TYPES_AND_HALF

* revert scalar_type() to type() in AT_DISPATCH_FLOATING_TYPES

* revert scalar_type() to type() for AT_DISPATCH_FLOATING_TYPES_AND_HALF in welford.cu

* revert scalar_type() to type() in layer_norm_cuda_kernel.cu

* revert at::kType  to at::ScalarType::Type

* use DISPATCH_FLOAT_AND_HALF to get rid of warnings

* add dispatch mechanisms for double+float and double+float+half
      855808f3
  15. 10 Apr, 2019 2 commits
  17. 09 Apr, 2019 1 commit
  19. 08 Apr, 2019 1 commit
  21. 04 Apr, 2019 1 commit
    • mcarilli's avatar
      WIP: Handle arbitrary combinations of optimizers/models/losses (#232) · 3f87614f
      mcarilli authored 
      * Refactor to allow more flexible treatment of multiple optimizers/models/losses

* Adding _process_optimizers.py

* Created L0 tests (now passing).

* fix: minor print typo (#234)

* make L1 results easier to read

* L0 multiple model/optimizer/loss test fleshed out

* Adding test that master params remain synced across distributed processes

* Docstring updates

* Docstring updates
      3f87614f
  23. 21 Mar, 2019 2 commits
  25. 19 Mar, 2019 2 commits
  27. 15 Mar, 2019 1 commit
  29. 12 Mar, 2019 1 commit
  31. 10 Mar, 2019 2 commits
  33. 03 Mar, 2019 1 commit
  35. 28 Feb, 2019 1 commit
  37. 24 Feb, 2019 1 commit
  39. 22 Feb, 2019 1 commit
  41. 19 Feb, 2019 1 commit
  43. 13 Feb, 2019 1 commit
  45. 11 Feb, 2019 1 commit
  47. 08 Feb, 2019 1 commit
  49. 06 Feb, 2019 2 commits
  51. 05 Feb, 2019 1 commit
  53. 04 Feb, 2019 1 commit
  55. 01 Feb, 2019 1 commit
  57. 18 Jan, 2019 1 commit
  59. 15 Jan, 2019 1 commit
    • Jie's avatar
      [sync BN nhwc] · 443fa76e
      Jie authored 
      Added kernel to support sync BN for channel last tensor
      443fa76e
  61. 06 Nov, 2018 1 commit
    • Jie's avatar
      [syncBN] · ee67e56a
      Jie authored 
      adjusted kernel config for better perf.
removed divergence in welford warp reduction.
      ee67e56a
  63. 30 Oct, 2018 1 commit
  65. 29 Oct, 2018 1 commit
    • mcarilli's avatar
      Merging in fused adam optimizer, additional DDP features tested in 18.10 (#60) · e0bc5d62
      mcarilli authored 
      * test passes

* notes

* Using C++-side flatten and unflatten functions

* Adding csrc

* Persistent synchronization event so it doesn't need to be created and destroyed each time

* Interop with parameter flattening in SSD

* Added deterministic option to imagenet main.py

* Adding options to split gradient averaging and allreduce in pure fp32

* Fixing allreduce_maybe_retain call

* Fixing allreduce_fallback

* Also sync active_i_buckets from rank 0

* Making retain_allreduce_buffers compatible with/orthogonal to delay_allreduce=True|False

* Correcting syntax error, now all seems to work with SSD

* Optional cpp extension build

* Add mixed precision adam optimizer (#59)

* Add FusedAdam Optimizer to Apex that places all the math into a cuda kernel.

* Added fixes to fused_adam to get it to work with network.

* wip work on python interface for adam with options

* fix dispatch for halfs, add python options to handle optional half gradients and params

* cleanup, get rid of grid-stride loop
      e0bc5d62