What's the license of this dataset?
#3
by
upstartupstart
- opened
Thanks for making this!
I'd love to use this, but I notice that there's no license...could you clarify whether this dataset is open to use under what conditions?
Hey @upstartupstart , thanks for your interest! There are two parts to this dataset:
- "categories" and "emojis" subsets, which are collected and annotated by me. The license for those annotations is CC BY-NC SA 4.0 https://creativecommons.org/licenses/by-nc-sa/4.0/.
- Synthetically generated labels for emojis and their embeddings. Those depend on the LLM that was used to generate the caption, and the embeddings model for the embeddings. Splits that end with "madmon" and "madmon-medium" are embedded with our proprietary model and also licensed as CC BY-NC SA 4.0.
Some of the content uses its original license. For example several emoji names were obtained using the emoji python package (see its supported languages section) which repackages Unicode specifications which are published under https://opensource.org/license/unicode-license-v3.
Sorry for the complex answer! I'm guessing you are probably fine for non-commercial use with attribution.