Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
Users might not care about the autofilled metrics. To improve the readability, we could add a feature that disables autofilled metrics.
train.report(epoch=1)
# {'epoch': 1}
Use case
Adding this feature could make reported data easier to work with.
Additionally, disabling autofilled metrics could improve the quality of our tests. Right now, tests are complicated as to handle the autofilled metrics.
assert all(
all(all(key in worker for key in BASIC_AUTOFILLED_KEYS) for worker in element)
for element in log
)
Search before asking
Description
Currently, Train reports are autofilled with various metrics.
Users might not care about the autofilled metrics. To improve the readability, we could add a feature that disables autofilled metrics.
Use case
Adding this feature could make reported data easier to work with.
Additionally, disabling autofilled metrics could improve the quality of our tests. Right now, tests are complicated as to handle the autofilled metrics.
Related issues
No response
Are you willing to submit a PR?