Skip to content

v0.4.0rc5

Pre-release
Pre-release
Compare
Choose a tag to compare
@github-actions github-actions released this 20 Jan 02:14
· 504 commits to main since this release
v0.4.0rc5
42654e6

What's new

Added 🎉

  • Added TorchEvalStep to torch integration, registered as "torch::eval".

Changed ⚠️

  • Renamed aggregate_val_metric to auto_aggregate_val_metric in TorchTrainStep.
  • devices parameter to TorchTrainStep replaced with device_count: int.
  • Run name printed at the end of a run so it's easier to find.
  • Type information added to package data. See PEP 561 for more information.
  • A new integration, transformers, with two new steps for running seq2seq models.
  • Added logging_tqdm, if you don't want a progress bar, but you still want to see progress in the logs.
  • Added threaded_generator(), for wrapping generators so that they run in a separate thread from the generator's consumer.
  • Added a new example for evaluating the T0 model on XSum, a summarization task.
  • Added MappedSequence for functionally wrapping sequences.
  • Added TextFormat, in case you want to store the output of your steps in raw text instead of JSON.
  • Steps can now list arguments in SKIP_ID_ARGUMENTS to indicate that the argument should not affect a step's
    unique id. This is useful for arguments that affect the execution of a step, but not the output.
  • Step now implements __str__, so steps look pretty in the debugger.
  • Added DatasetCombineStep, a step that combines multiple datasets into one.
  • Added common.logging.initialize_worker_logging() function for configuring logging from worker processes/threads.
  • Logs from tango run ... will be written to a file called out.log in the run directory.

Fixed ✅

  • Fixed torch StopEarlyCallback state not being recovered properly on restarts.
  • Fixed file friendly logging by removing special styling characters.
  • Ensured exceptions captured in logs.
  • LocalWorkspace now works properly with uncacheable steps.
  • When a Tango run got killed hard, with kill -9, or because the machine lost power, LocalWorkspace would
    sometimes keep a step marked as "running", preventing further executions. This still happens sometimes, but it
    is now much less likely (and Tango gives you instructions for how to fix it).
  • To make all this happen, LocalWorkspace now saves step info in a Sqlite database. Unfortunately that means that
    the workspace format changes and existing workspace directories won't work properly with it.
  • Fixed premature cleanup of temporary directories when using MemoryWorkspace

Commits

df301ef Merge pull request #119 from allenai/RunGeneration
42535c9 Add TorchEvalStep to torch integration, use "device_count" in TorchTrainStep instead of "devices" (#120)
ecc6087 Store log output to a file in run directory, other logging improvements (#132)
9008255 Merge pull request #141 from allenai/dependabot/pip/sphinx-4.4.0
8e09b66 Merge pull request #139 from allenai/remove-no-logging
26318b2 Merge pull request #133 from allenai/dependabot/pip/mypy-0.931
7c91c4b Merge pull request #126 from allenai/dependabot/pip/sphinx-4.3.2
a178076 Merge pull request #128 from allenai/dependabot/pip/mypy-0.930
106a8cf Merge pull request #130 from allenai/dependabot/pip/furo-2022.1.2
6db7b4c Merge pull request #129 from allenai/run-name-at-end
b93d22d Add typing info to package (PEP 561) (#131)
e8867ae fix StopEarlyCallback state recovery

Assets 4
Loading