Datasets
Standard Dataset
AGI language training
- Citation Author(s):
- Submitted by:
- Valery Shmeleff
- Last updated:
- Wed, 01/01/2025 - 07:49
- DOI:
- 10.21227/bdw2-dr89
- Data Format:
- License:
- Categories:
- Keywords:
Abstract
Memes (photos with text) for fine-tuning AGI. Training in recognizing and generating memes, jokes, mockery, trolling, emotions in foreign languages. This is a demo sample (will be expanded) before data expansion. Hundreds of variants were created for each meme. Meme text without using the expanded data set is difficult enough for AGI to correctly recognize emotions. With the data set expansion, the accuracy increases significantly.
Correct recognition of memes and jokes in foreign languages has many difficulties associated with the emergence of new jokes that have not previously been used in training, the influence of intonation during pronunciation (pictures with text must be voiced by a qualified specialist).
This is a very special Data Set for training the recognition and generation of emotions.
Documentation
Attachment | Size |
---|---|
About Data Set | 781 bytes |