- Dan Saattrup Nielsen (@saattrupdan, [email protected])
- Kenneth Enevoldsen (@KennethEnevoldsen, [email protected])
To install the package simply write the following command in your favorite terminal:
$ pip install scandeval[all]
This will install the ScandEval package with all extras. You can also install the
minimal version by leaving out the [all]
, in which case the package will let you know
when an evaluation requires a certain extra dependency, and how you install it.
The easiest way to benchmark pretrained models is via the command line interface. After having installed the package, you can benchmark your favorite model like so:
$ scandeval --model <model-id>
Here model
is the HuggingFace model ID, which can be found on the HuggingFace
Hub. By default this will benchmark the model on all
the tasks available. If you want to benchmark on a particular task, then use the
--task
argument:
$ scandeval --model <model-id> --task sentiment-classification
We can also narrow down which languages we would like to benchmark on. This can be done
by setting the --language
argument. Here we thus benchmark the model on the Danish
sentiment classification task:
$ scandeval --model <model-id> --task sentiment-classification --language da
Multiple models, datasets and/or languages can be specified by just attaching multiple arguments. Here is an example with two models:
$ scandeval --model <model-id1> --model <model-id2>
The specific model version/revision to use can also be added after the suffix '@':
$ scandeval --model <model-id>@<commit>
This can be a branch name, a tag name, or a commit id. It defaults to 'main' for latest.
See all the arguments and options available for the scandeval
command by typing
$ scandeval --help
In a script, the syntax is similar to the command line interface. You simply initialise
an object of the Benchmarker
class, and call this benchmark object with your favorite
model:
>>> from scandeval import Benchmarker
>>> benchmark = Benchmarker()
>>> benchmark(model="<model>")
To benchmark on a specific task and/or language, you simply specify the task
or
language
arguments, shown here with same example as above:
>>> benchmark(model="<model>", task="sentiment-classification", language="da")
If you want to benchmark a subset of all the models on the Hugging Face Hub, you can
simply leave out the model
argument. In this example, we're benchmarking all Danish
models on the Danish sentiment classification task:
>>> benchmark(task="sentiment-classification", language="da")
A Dockerfile is provided in the repo, which can be downloaded and run, without needing to clone the repo and installing from source. This can be fetched programmatically by running the following:
$ wget https://raw.githubusercontent.com/ScandEval/ScandEval/main/Dockerfile.cuda
Next, to be able to build the Docker image, first ensure that the NVIDIA Container
Toolkit is
installed
and
configured.
Ensure that the the CUDA version stated at the top of the Dockerfile matches the CUDA
version installed (which you can check using nvidia-smi
). After that, we build the
image as follows:
$ docker build --pull -t scandeval -f Dockerfile.cuda .
With the Docker image built, we can now evaluate any model as follows:
$ docker run -e args="<scandeval-arguments>" --gpus 1 --name scandeval --rm scandeval
Here <scandeval-arguments>
consists of the arguments added to the scandeval
CLI
argument. This could for instance be --model <model-id> --task sentiment-classification
.
- Thanks to OpenAI for sponsoring OpenAI credits as part of their Researcher Access Program.
- Thanks to UWV and KU Leuven for sponsoring the Azure OpenAI credits used to evaluate GPT-4-turbo in Dutch.
- Thanks to Miรฐeind for sponsoring the OpenAI credits used to evaluate GPT-4-turbo in Icelandic and Faroese.
- Thanks to CHC for sponsoring the OpenAI credits used to evaluate GPT-4-turbo in German.
If you want to cite the framework then feel free to use this:
@article{nielsen2024encoder,
title={Encoder vs Decoder: Comparative Analysis of Encoder and Decoder Language Models on Multilingual NLU Tasks},
author={Nielsen, Dan Saattrup and Enevoldsen, Kenneth and Schneider-Kamp, Peter},
journal={arXiv preprint arXiv:2406.13469},
year={2024}
}
@inproceedings{nielsen2023scandeval,
author = {Nielsen, Dan Saattrup},
booktitle = {Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa)},
month = may,
pages = {185--201},
title = {{ScandEval: A Benchmark for Scandinavian Natural Language Processing}},
year = {2023}
}
The image used in the logo has been created by the amazing Scandinavia and the World team. Go check them out!