search

hrshtv / CTreeOT

[ICML 2023] Official code for our paper: 'Conditional Tree Matching for Inference-Time Adaptation of Tree Prediction Models'
4 stars 0 forks source link

Request for Exact Data Splits Used in Adaptation to SPIDER Schemas #1

Open llm-based-developement opened 3 months ago

llm-based-developement commented 3 months ago

Thanks a lot for your work.

I am currently trying to replicate the experiments you've conducted. In the paper, you mentioned that divisions of the schema D were performed randomly into Dtest and Dcases such that Dtest ∪ Dcases = D and Dtest ∩ Dcases = ∅

I understand that based on random splits, the results could vary. For comparing different text-to-SQL work, it would be ideal to keep the experiment setting similar.

Therefore, it would be really helpful if you could please share the exact Data splits used in your experiments.

hrshtv commented 2 months ago

Hi, we use the same data splits as StructCBR (https://github.com/awasthiabhijeet/structcbr).