Hugging Face's logo

atlasia
/
Terjman-Supreme

Text2Text Generation
Transformers
Safetensors
Arabic
English
m2m_100
Inference Endpoints
Model card Files Community
1

Evaluation Set

#1
by Aiyagh - opened
Discussion
Aiyagh

Hey, thank you for your work and this model. I was just wondering what evaluation set are you using since you are mentioning the following in the model card.

It achieves the following results on the evaluation set:

Loss: 2.3687
Bleu: 5.6718
Gen Len: 39.9504
BounharAbdelaziz
 AtlasIA org

Hi @Aiyagh , thanks for your feedback. The evaluation set is a subset of the training dataset "atlasia/darija_english" that the model hasn't seen during training :)

BounharAbdelaziz changed discussion status to closed
Aiyagh

Hi @BounharAbdelaziz , yeah, I get that, but I mean I looked at the dataset "atlasia/darija_english" and it only has a training split and no evaluation split or is the evaluation set private?

imomayiz
 AtlasIA org

@Aiyagh both the training and evaluation datasets are private. "atlasia/darija_english" is just a collection of curated datasets that includes a subset of the training data.

Aiyagh

@imomayiz Ah okay I see, thank you for the clarification.

Sign up or log in to comment