The AI developer platform. Use Weights & Biases to train and fine-tune models, and manage models from experimentation to production.
MIT License
Bot releases are visible (Hide)
Published by raubitsj over 2 years ago
args_no_boolean_flags
macro to sweep configuration by @hu-po in https://github.com/wandb/client/pull/3489
jax.bfloat.bfloat16
by @dmitryduev in https://github.com/wandb/client/pull/3528
wandb.init
is interrupted mid setup using wandb-service by @kptkin in https://github.com/wandb/client/pull/3569
wandb.init(settings=)
to handle Settings
object similarly to dict
parameter by @dmitryduev in https://github.com/wandb/client/pull/3510
Full Changelog: https://github.com/wandb/client/compare/v0.12.15...v0.12.16
Published by raubitsj over 2 years ago
Full Changelog: https://github.com/wandb/client/compare/v0.12.14...v0.12.15
Published by raubitsj over 2 years ago
Full Changelog: https://github.com/wandb/client/compare/v0.12.13...v0.12.14
Published by raubitsj over 2 years ago
Full Changelog: https://github.com/wandb/client/compare/v0.12.12...v0.12.13
Published by raubitsj over 2 years ago
Full Changelog: https://github.com/wandb/client/compare/v0.12.11...v0.12.12
Published by raubitsj over 2 years ago
wandb_callback
(#2929) by @ayulockin in https://github.com/wandb/client/pull/3025
sync_tensorboard
could die on large histograms by @KyleGoyette in https://github.com/wandb/client/pull/3019
Full Changelog: https://github.com/wandb/client/compare/v0.12.10...v0.12.11
Published by raubitsj over 2 years ago
use_artifact()
calls by @annirudh in https://github.com/wandb/client/pull/3109
Full Changelog: https://github.com/wandb/client/compare/v0.12.9...v0.12.10
Published by raubitsj almost 3 years ago
upload_file()
exception handler by @raubitsj in https://github.com/wandb/client/pull/3059
Full Changelog: https://github.com/wandb/client/compare/v0.12.8...v0.12.9
Published by raubitsj almost 3 years ago
wandb_callback
for LightGBM (#2945) by @ayulockin in https://github.com/wandb/client/pull/3024
from wandb import magic
to not require tensorflow by @raubitsj in https://github.com/wandb/client/pull/3021
Full Changelog: https://github.com/wandb/client/compare/v0.12.7...v0.12.8
Published by raubitsj almost 3 years ago
Full Changelog: https://github.com/wandb/client/compare/v0.12.6...v0.12.7
Published by raubitsj almost 3 years ago
plot_calibration_curve()
issue breaking the provided model by @vwrj in https://github.com/wandb/client/pull/2791
use_artifact()
when specifying an artifact from a different project by @KyleGoyette in https://github.com/wandb/client/pull/2832
Full Changelog: https://github.com/wandb/client/compare/v0.12.5...v0.12.6
Published by raubitsj almost 3 years ago
%%wandb
followed by wandb.init() does not display run linkswandb.require(experiment="service")
to improve multiprocessing supportwandb.login(timeout=)
support for jupyter environmentsWANDB_DISABLE_GIT
artifact.delete(delete_aliases=True)
unwatch()
method to the Run objectFull Changelog: https://github.com/wandb/client/compare/v0.12.4...v0.12.5
Published by raubitsj about 3 years ago
WANDB_MODE=offline
Full Changelog: https://github.com/wandb/client/compare/v0.12.3...v0.12.4
Published by raubitsj about 3 years ago
%wandb path/to/run -h 1024
%%wandb
to a cell to display a runapi.team("team_name").invite("username_or_email")
api.team("team_name").members[0].delete()
api.team("team_name").create_service_account("Description")
api.user("username_or_email").generate_api_key()
wandb.profiler.torch_trace_handler()
Published by raubitsj about 3 years ago
wandb.login()
when relogin is specified (only force login once)wandb sweep --update
wandb local
users to update their dockerwandb launch
wandb.login()
timeout optionPublished by raubitsj about 3 years ago
WANDB_IGNORE_GLOBS
environment variable handlingPublished by raubitsj about 3 years ago
wandb.restore()
wouldn't work with runs from a sweepPublished by raubitsj about 3 years ago
Published by raubitsj about 3 years ago
wandb==0.12.0
pip install --upgrade urllib3
Published by dannygoldstein over 3 years ago
wandb.watch()
broke model saving in pytorch