-
-
Notifications
You must be signed in to change notification settings - Fork 7
Open
Labels
enhancementNew feature or requestNew feature or requestpipeline 5: testIssue relating to testing a model quality with Bleu or other metrics.Issue relating to testing a model quality with Bleu or other metrics.
Description
We want to save the scores (including confidence if it's being saved) for test.py at a verse-level automatically, rather than having to run diff_predictions afterward. Currently only It can be similar to how write_sentence_bleu works, but saving all the scorers used not just sentence_bleu. We can reuse the name that method uses: test.trg-predictions.detok.txt.[ckpt].scores.tsv.
Once that's done, we also need to change the quality_estimation script so that it takes in this file rather than diff_predictions as an input.
Metadata
Metadata
Assignees
Labels
enhancementNew feature or requestNew feature or requestpipeline 5: testIssue relating to testing a model quality with Bleu or other metrics.Issue relating to testing a model quality with Bleu or other metrics.
Type
Projects
Status
🔖 Ready