Hugging Face's logo

HuggingFaceTB
/
SmolLM2-135M

Text Generation
Transformers
Safetensors
English
llama
text-generation-inference
Inference Endpoints
Model card Files Community
5

What MMLU (cloze) used for evaluation.

#2
by akhilfau - opened
Discussion
akhilfau

Did you guys used below dataset or some other one from hugging face.
https://huggingface.co/datasets/cais/mmlu

loubnabnl
 Hugging Face TB Research org

We use this dataset https://huggingface.co/datasets/lighteval/mmlu (cf this code https://github.com/huggingface/cosmopedia/blob/653acad4f6146ce1043e2ca8792c671947e83ce0/evaluation/lighteval_tasks.py#L280)

Sign up or log in to comment