3 ms·
Can someone provide the prompts in text rather than the images from the paper? That would make it easier to try and replicate results.
by kenjackson 3y ago
Can someone provide the prompts in text rather than the images from the paper? That would make it easier to try and replicate results.
- devbrac 3y agoRaw json files for each task from the ConceptARC paper are here: https://github.com/victorvikram/ConceptARC https://github.com/victorvikram/ConceptARC. It would be trivial to convert these into prompts using the format in the appendix of the paper.