mila-iqia/milabench

Issues

• rank 0 is not on the local node

  #282 opened 4 months ago by Delaunay
  2

• diffusion-nodes can fail loading the checkpoint on some rank

  #281 opened 4 months ago by Delaunay
  6

• Jax & CUDA

  #258 opened 4 months ago by Delaunay
  0

• amdsmi not found

  #256 opened 4 months ago by Delaunay
  0

• Use llama3 for inference

  #255 opened 4 months ago by Delaunay
  0

• Milabench Size

  #254 opened 4 months ago by Delaunay
  0

• light docker image

  #244 opened 5 months ago by Delaunay
  0

• Data pack redistribution

  #239 opened 5 months ago by Delaunay
  0

• ptera crashes on Gracehooper

  #216 opened 6 months ago by Delaunay
  5

• Install & Prepare should work without GPU

  #226 opened 6 months ago by Delaunay
  1

• Document batch resizing

  #211 opened 9 months ago by Delaunay
  0

• v0.0.7 master branch: local variable 'timeout_task' referenced before assignment

  #164 opened a year ago by luop0812
  3

• NonMatchingSplitsSizesError on Prepare Step

  #184 opened a year ago by ArjunRamaswami22
  4

• Failures when running on a 4 GPU server

  #183 opened a year ago by guylaporte
  4

• CUDA out of memory

  #192 opened a year ago by lcebaman
  3

• Validation Layers

  #57 opened 2 years ago by Delaunay
  0

• `milabench --help` raise

  #103 opened 2 years ago by Delaunay
  0

• AMD: Use all available GPUs for multi-gpu tests

  #46 opened 2 years ago by Delaunay
  1

• RL benchmarks download atari ROM from website that are often behind company firewall

  #69 opened 2 years ago by Delaunay
  1

• vit_l_32 fails on H100 with pytroch nightly

  #70 opened 2 years ago by Delaunay
  1

• use https instead of git uris

  #4 opened 2 years ago by tbugfinder
  6

• move anaconda installation

  #5 opened 2 years ago by tbugfinder
  2

• scaling benchmark fails

  #1 opened 2 years ago by nileshnegi
  1

• Error running milabench with H100 GPU

  #44 opened 2 years ago by guylaporte
  18

• Allow to override dependencies at the top level

  #20 opened 2 years ago by Delaunay
  2

• Interrupting milabench creates dangling `voir` processes

  #15 opened 2 years ago by lebrice
  0

• provide prebuilt docker image

  #2 opened 2 years ago by tbugfinder
  1

• Investigate Batch size scaling in DDP setups

  #60 opened 2 years ago by Delaunay
  0

• Add a flag to make milabench exit with -1 to make tests fails when something is wrong

  #54 opened 2 years ago by Delaunay
  1

• Install crash with newest version of pytorch bench

  #29 opened 2 years ago by Delaunay
  5

• Milabench v2 on AMD

  #21 opened 2 years ago by Delaunay
  8

• Add AMD GPU monitoring

  #22 opened 2 years ago by Delaunay
  0

• milabench install requires GPU

  #42 opened 2 years ago by Delaunay
  3

• Problem installing milabench

  #41 opened 2 years ago by guylaporte
  5

• MILABENCH_GPU_ARCH undocumented

  #39 opened 2 years ago by gravitino
  2

• Use NVML instead of parsing nvidia-smi output

  #38 opened 2 years ago by gravitino
  2

• numpy version requirement problem with torchbench

  #32 opened 2 years ago by abergeron
  1

• soft_actor_critic crash because of API change

  #28 opened 2 years ago by Delaunay
  2

• soft_actor_critic crash because of API change

  #27 opened 2 years ago by Delaunay
  0

• [low] ptera.selfless.ConflictError: Multiple values with same priority conflict for variable 'base'

  #9 opened 3 years ago by lebrice
  3

• Create license

  #6 opened 4 years ago by fosterrath-mila
  1

• build using podman/buildah

  #3 opened 4 years ago by tbugfinder
  1