Labels

• activation checkpoint
• AdaScale

  Optimizer wrapper for automatic LR scaling without lossing model accuracy
• better_eng
• bug

  Something isn't working
• CLA Signed

  This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
• dependencies

  Pull requests that update a dependency file
• documentation

  Improvements or additions to documentation
• duplicate

  This issue or pull request already exists
• enhancement

  New feature or request
• FSDP

  FullyShardedDataParallel (zero-3)
• FSDP + SSD offload
• good first issue

  Good for newcomers
• help wanted

  Extra attention is needed
• icebox
• in_progress

  this issue is being worked on
• invalid

  This doesn't seem right
• MEVO

  memory efficient vocab output
• moe
• offload_model
• OSS

  Optimizer State Sharding (zero-1)
• Pipeline

  Pipeline parallelism
• question

  Further information is requested
• SDP

  ShardedDataParallel (zero-2)
• triaged
• unit test
• wontfix

  This will not be worked on