unitxt
unitxt copied to clipboard
🦄 Unitxt: a python library for getting data fired up and set for training and evaluation
Adding support for llama-3-3-70b model from WML.
closes https://github.com/IBM/unitxt/issues/1894 and also fixes small issues that prevent the recipes generated from the bfcl cards prepared to be well processed by `_source_to_dataset` (mainly schema issues) Errors before the fix:...
There are 30 cards in the group `cards.mmmu.*`, 16 of which (more than half) are erroneous: do not pass `unitxt.api.load_dataset` : [mmmu_main.pdf](https://github.com/user-attachments/files/21888780/mmmu_main.pdf) Exploring the original HF datasets, the following came...
remove `register_all_artifacts` from `unitxt.__init__`. Instantiate classes upon demand (only the needed ones) by finding the module, and then use import tools.
**BEFORE:** [error_box_main.pdf](https://github.com/user-attachments/files/22126122/error_box_main.pdf) **AFTER:** [error_box_fixed.pdf](https://github.com/user-attachments/files/22127228/error_box_fixed.pdf)
Welcome to Mend for GitHub Enterprise! This is an onboarding PR to help you understand and configure settings before Mend starts scanning your repository for security vulnerabilities. :vertical_traffic_light: Mend for...
If you cannot find operators fit to your needs simply use function to modify every instance in the data: ```python def my_function(instance, stream_name=None): instance["x"] += 42 return instance ``` Or...