Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Answered
Tracking From Experiments To Datasets

Tracking from Experiments to Datasets

Hi all - is there an easier way track the set of datasets used by a particular task? Consumed models show up as "Input Models" in the "Artifacts" tab, with a nice link to the model and a "creating experiment" link to trace back to the task which created the model. I was hoping for the same thing for Datasets, but all I get is the IDs in the task's "Info" panel. Worse, if I copy-paste the ID and try to search for it, I get no hits, neither in the global search context on the home page, nor in the dataset search context on the Datasets page. I did confirm that this is indeed the correct dataset ID by manually navigating to the dataset.

Any thoughts?

  
  
Posted one year ago
Votes Newest

Answers 6


Oh, I was more referring to having a clickable link for the dataset, rather than adding the dataset GUID to a search index.

  
  
Posted one year ago

Both fixes would improve things, but making the dataset clickable -- similar to how Input Model is clickable -- would be best.

  
  
Posted one year ago

Fixing that would make this feature great.

Hmm, I guess that is doable, this is a good point, search for the GUID is not always trivial (or maybe at least we can put in the description the project/dataset/version )

  
  
Posted one year ago

Hi AmiableFish73

Hi all - is there an easier way track the set of datasets used by a particular task?

I think the easiest is to give the Dataset an alias, it will automatically appear in the Configuration section:
Dataset.get(..., alias="train dataset")wdyt?

  
  
Posted one year ago

Yeah that make sense 🙂

  
  
Posted one year ago

Hi @<1523701205467926528:profile|AgitatedDove14> - yes, that's great! It's much cleaner. I was thinking of tracking the input datasets as artifacts, but the alias method is better.

The user interface still only displays a GUID - it would be much preferrable if it displayed a clickable link which took me to the dataset details page -- particularly since the GUID doesn't search properly.

Fixing that would make this feature great.

  
  
Posted one year ago