ServiceNow/azimuth
Helping AI practitioners better understand their datasets and models in text classification. From ServiceNow.
PythonApache-2.0
Issues
- 0
Status checks do not run for PR submitted from forks
#621 opened by istandre - 6
Docker build fails
#619 opened by WiseBird - 2
- 2
Auto selection of the saliency layer
#582 opened by Dref360 - 0
- 0
Utterance details as a modal.
#555 opened by gabegma - 2
Investigate saliency task - is memory leaking?
#549 opened by gabegma - 1
Keep GPU workload to one worker locally.
#411 opened by Dref360 - 0
Tasks often get lost because a worker dies.
#521 opened by gabegma - 0
Debug logs are not showing on Dask workers.
#548 opened by gabegma - 0
- 0
Limit the number of DatasetSplitManager and Model that the ArtifactManager can have in memory
#531 opened by gabegma - 0
Run the startup tasks per pipeline separately to avoid loading multiple pipelines at the same time.
#530 opened by gabegma - 0
Make config checkboxes' "hit box" smaller
#488 opened by JosephMarinier - 0
Outcome per threshold chart overflow the tile
#483 opened by gabegma - 0
Make smart tag aggregation plot work for dataset only
#433 opened by gabegma - 0
Ability to import back proposed actions
#297 opened by gabegma - 3
Tensor size mismatch with HF pipeline
#439 opened by vilhub - 0
App don't load when there is no training set
#358 opened by gabegma - 0
- 0
Handle `"pipelines": []` in config
#335 opened by JosephMarinier - 0
Convert short_sentence and long_sentence to short_utterance and long_utterance.
#434 opened by gabegma - 1
Use the confusion matrix as a filter
#469 opened by JosephMarinier - 0
Update datasets to v2.1.0
#267 opened by gabegma - 1
Show persistent id in front end
#377 opened by JosephMarinier - 0
Make Azimuth work on just a training set
#362 opened by gabegma - 0
Refactor get_utterances so it runs faster
#451 opened by gabegma - 0
When modifying the config to a previous value (where a module cache exists), the dataset cache won't match.
#453 opened by gabegma - 0
Add Outcomes to dropdown in Smart Tag Analysis
#431 opened by gabegma - 0
Change the outcome per threshold bar plot to area plot and add the current threshold value as a vertical line
#430 opened by gabegma - 0
Add option to show the smart tag aggregation plot by smart tag values instead of families
#432 opened by gabegma - 0
- 1
- 0
Refactor TaskManager so it doesn't use config as an attribute, and doesn't need to be killed.
#347 opened by gabegma - 0
Save config files in caching folder with json lines
#345 opened by gabegma - 0
- 0
- 0
Sorting is using the post-processed confidence even when the toggle for removing post-processing is activated.
#319 opened by gabegma - 0
In the utterance details, the modified utterance is not readable when the utterance is long.
#320 opened by gabegma - 1
Web app no longer loading when there is no pipeline
#331 opened by gabegma - 0
Overlap threshold is not easy to change.
#294 opened by gabegma - 0
Detect when module options are not valid
#305 opened by gabegma - 0
Determine good min and max for class overlap slider
#308 opened by gabegma - 0
Bug in class analysis -> pipeline confusion all at 0
#303 opened by gabegma - 0
Show the details of the preprocessing and postprocessing steps in the utterance details.
#283 opened by gabegma - 0
Add accuracy
#295 opened by gabegma - 0
Update spaCy (and models) to 3.4.1 and test
#280 opened by lindsaydbrin - 1
Ability to block the modification of the config
#268 opened by gabegma - 0
- 0
Add class imbalance warnings
#271 opened by gabegma