GSM8k) a dataset of 8.5K high quality linguistically diverse grade school math word problems.
The GSM8k benchmark has been translated into Lithuanian using GPT-4. This dataset is utilized as a benchmark and forms part of the evaluation protocol for Lithuanian language models, as outlined in the technical report OPEN LLAMA2 MODEL FOR THE LITHUANIAN LANGUAGE (Nakvosas et al., 2024)
@article{cobbe2021trainingverifierssolvemath… See the full description on the dataset page:
https://huggingface.co/datasets/neurotechnology/lt_gsm8k.
