Statements (50)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
model
|
| gptkbp:architect |
transformer
|
| gptkbp:completed |
state-of-the-art performance
|
| gptkbp:developedBy |
OpenAI
|
| gptkbp:hasCitations |
bias in training data
generalization to unseen data interpretability issues |
| gptkbp:hasImpactOn |
AI safety
AI ethics |
| gptkbp:hasRelatedPatent |
computer vision
natural language processing content moderation art generation image search text search |
| https://www.w3.org/2000/01/rdf-schema#label |
CLIP
|
| gptkbp:influencedBy |
GPT-3
|
| gptkbp:inputOutput |
text
images |
| gptkbp:isAvailableIn |
GitHub
Hugging Face |
| gptkbp:isCitedBy |
industry reports
numerous academic papers AI conferences |
| gptkbp:isCompatibleWith |
gptkb:PyTorch
TensorFlow |
| gptkbp:isPartOf |
AI research
OpenAI's research portfolio |
| gptkbp:isRelatedTo |
gptkb:VQGAN
BERT StyleGAN DALL-E Vision Transformers |
| gptkbp:isSuitableFor |
yes
|
| gptkbp:isUsedBy |
artists
developers researchers |
| gptkbp:mayHave |
image generation
zero-shot classification image-to-text retrieval text-to-image retrieval |
| gptkbp:powerOutput |
image embeddings
text embeddings |
| gptkbp:relatedTo |
traditional supervised learning models
|
| gptkbp:releaseYear |
2021
|
| gptkbp:requires |
large datasets
computational resources |
| gptkbp:training |
image-text pairs
|
| gptkbp:uses |
multi-modal learning
contrastive learning |