All pull requests

Pull requests

• Fix wds2idx offsets of members with GNU long names

  #6547·MaxFreedomPollard opened Oct 22, 2026Contributor·3 of 18 tasks·

  ·
  2

• Fix GCC 11 warning in color conversion tests

  #6546·JanuszL opened Oct 22, 2026Contributor·3 of 18 tasks·

  ·
  4

• Decode the video reader's batches in parallel with decode lanes

  #6545·jantonguirao opened Oct 22, 2026Collaborator·7 of 18 tasks·

  ·
  11

• [Torchvision API] Support for volumetric data

  #6544·mdabek-nvidia opened Oct 22, 2026Collaborator·4 of 18 tasks·

  ·
  7

• Add any and all functions to dynamic mode. Support bool in min and max reductions.

  dynamic mode
  #6543·mzient opened Oct 22, 2026Collaborator·6 of 18 tasks·

  ·
  22

  

• Reject only overflow for float labels in readers.file

  #6542·jantonguirao opened Oct 22, 2026Collaborator·5 of 18 tasks·

  ·
  16

  

• Add a benchmark and a TL1 perf test for fn.experimental.readers.video

  #6541·jantonguirao opened Oct 21, 2026Collaborator·3 of 18 tasks·

  ·
  56

  

• Fix NVDEC decoder cache never matching cropped videos

  #6540·jantonguirao opened Oct 21, 2026Collaborator·4 of 18 tasks·

  ·
  9

  

• Fix Python sanitizer test failures

  #6539·JanuszL opened Oct 21, 2026Contributor·7 of 18 tasks·

  ·
  21

  

• Keep trailing dims in the layout of an indexed dynamic Tensor

  #6538·v0ropaev opened Oct 21, 2026Contributor·3 of 18 tasks·

  ·
  2

  

• Fix sample selection in Batch.tensors and Batch.select

  #6537·v0ropaev opened Oct 21, 2026Contributor·3 of 18 tasks·

  ·
  4

  

• Fix float16 to integer normalized conversion and convert.h doc typos

  #6535·jantonguirao opened Oct 20, 2026Collaborator·5 of 18 tasks·

  ·
  3

  

• Update nvCOMP to 6.0 for CUDA 12.9 and 13.4 builds

  #6530·JanuszL opened Oct 19, 2026Contributor·5 of 18 tasks·

  ·
  18

  

• Handle spurious device errors in capture mode

  dynamic modetrivial change
  #6528·rostan-t opened Oct 16, 2026Collaborator·4 of 18 tasks·

  ·
  5

  

• Support batch slicing in capture mode

  dynamic mode
  #6527·rostan-t opened Oct 15, 2026Collaborator·5 of 18 tasks·

  ·
  20

  

• Support arithmetic ops and math functions in capture mode

  dynamic mode
  #6525·rostan-t opened Oct 14, 2026Collaborator·5 of 18 tasks·

  ·
  27

  

• Enable GCS and S3 storage in video reader.

  #6518·mzient opened Oct 2, 2026Collaborator·4 of 18 tasks·

  ·
  21

  

• Reject an explicit zero float width instead of the IEEE default

  #6515·SashaMIT opened Sep 30, 2026·4 of 18 tasks·

  ·
  7

  

• Bring experimental.readers.video to feature parity with legacy readers.video

  #6513·jantonguirao opened Sep 29, 2026Collaborator·8 of 18 tasks·

  ·
  6

• Remove pickle from parallel external source shared memory messages

  #6512·devin-ai-integration[bot] opened Sep 29, 2026Bot·7 of 19 tasks·

  ·
  19

  

• Add keypoints output to fn.readers.coco

  #6509·devin-ai-integration[bot] opened Sep 29, 2026Bot·6 of 18 tasks·

  ·
  12

  

• Report which Python workers exited and why when worker communication fails

  #6508·devin-ai-integration[bot] opened Sep 29, 2026Bot·4 of 18 tasks·

  ·
  1

  13

  

• Merge uniform input/argument pairs in ndd, rename Clang build tag

  #6503·jantonguirao opened Sep 22, 2026Collaborator·7 of 18 tasks·

  ·
  41

  

• Fix fn API mentions leaking into deprecation warnings and docstrings

  #6502·jantonguirao opened Sep 22, 2026Collaborator·4 of 18 tasks·

  ·
  18

  

• Adding pip requirements.txt required for DALI compilation

  #6501·MykolaSemylitko opened Sep 22, 2026Collaborator·3 of 18 tasks·

  ·
  1