unitxt
unitxt copied to clipboard
🦄 Unitxt: a python library for getting data fired up and set for training and evaluation
When using HF APIs only the HF version code and catalog will be used
Following the Rouge updates , there are changes in rouge scores. In most cases, the diff is less than 1 point, , but it could be 1-2 point in extreme...
Right now , metrics score overwrite each other (Not referring to the 'score' and 'score_name' , but two metrics can have the same 'f1' score. Users find it confusing. We...
Currently, the full clapNQ dataset contains thousands of documents. It is not usable for testing a simple rag end-to-end flow. we would like to create a subset version of these...
I created a branch: and PR in draft state - https://github.com/IBM/unitxt/pull/953 * The task is there, but it fails to run because rouge is missing. I believe the metric is...
Also, we don't see the print out to see what takes the time. https://github.com/IBM/unitxt/actions/runs/10042751612/job/27753878493?pr=1034
currently we pass data_classification_policy in 2 places in the recipe As a field and in the task_data. We need to think if and how to consolidate them. for example: ```...