Ville Tuulos
Ville Tuulos
another 👍 https://gitter.im/metaflow_org/community?at=6081a180b9e6de24d64fb32e
another +1 here https://gitter.im/metaflow_org/community?at=6075e54e06e2e024e84f645c
another motivating example by @hugobowne using XGBoost `DMatrix`: ``` Metaflow 2.5.0 executing ClassificationFlow for user:hba Validating your flow... The graph looks good! Running pylint... Pylint is happy! 2022-02-24 15:59:14.529 Workflow...
this is particularly annoying when a long-running model training step fails wasting hours of work 
Yep, the topic probably deserves a design memo of its own but here's a quick strawman proposal to seed the discussion: 1. new command `metaflow garbage-collect --flow MyFlow` 2. the...
another thread for context and +1 https://outerbounds-community.slack.com/archives/C02116BBNTU/p1651716562792139
Good point. We should change this.
This is a tricky problem. Currently tdbcli makes no attempt in producing very "valid" CSV or JSON. Especially producing valid JSON is hard, since the RFC requires that JSON text...
duplicates in this context means that all fields are equal, including the timestamp and the uuid? Implementing dedup logic like this should be quite doable.
Interesting. If rpi had a 64-bit kernel, it shouldn't be a problem. TrailDB is not (well) supported on 32 bit platforms currently.