unitxt icon indicating copy to clipboard operation
unitxt copied to clipboard

🦄 Unitxt: a python library for getting data fired up and set for training and evaluation

Results 201 unitxt issues
Sort by recently updated
recently updated
newest added

for example when trying to create llm-as-judge using llama3 on hugging ace you will get the following error because you need to set an HF token ``` 403 Client Error....

bug

This would allow us an easier management and usage of templates and cards in 'produe'

Added a new metric using fuzzywuzzy logic for NER evaluation. This will help to compare the predictions with actual using string match based on predefined threshold.

Today, If metrics, returns the same score (e.g. "f1") then the first metric score is returned, and the second metric is ignored. metrics ["metrics.bert_score.distilbert_base_uncased","metrics.bert_score.deberta_base_mnli"], It would be good to be...

so the type checker knows that if we passed that function the type of the var is as mentioned. instanceof function have that effect.

enhancement

For example to support `List[Tuple[Literal["user", "assistant", "system"], str]] `

enhancement

For example, consider the empty format: ![image](https://github.com/IBM/unitxt/assets/68273864/2a66647e-2264-4dbc-8db3-f6c7ff66ee7e) Which today is: format = SystemFormat( demo_format="{source}\\N{target_prefix}{target}\n\n", model_input_format="{system_prompt}\\N{instruction}\\N{demos}{source}\\N{target_prefix}", )

Currently a specific template in unitxt have few versions for different languages. For example: English sentiment template: ```python template = InputOutputTemplate(input_format="Classify the sentiment of this text: {text}") ``` Deutch sentiment...

ease-of-use