PretrainingDataset
Enums · AiDotNet.VisionLanguage.Encoders
Specifies the pre-training dataset used for the model.
Fields
| Field | Summary |
|---|---|
PretrainingDataset CommonPool |
CommonPool filtered dataset. |
PretrainingDataset Custom |
Custom/other dataset. |
PretrainingDataset DFNFiltered |
DFN-filtered high-quality dataset. |
PretrainingDataset DataComp12B |
DataComp-12B dataset. |
PretrainingDataset DataComp1B |
DataComp-1B curated dataset. |
PretrainingDataset JFT300M |
JFT-300M Google internal dataset. |
PretrainingDataset LAION2B |
LAION-2B open dataset (2 billion image-text pairs). |
PretrainingDataset LAION400M |
LAION-400M open dataset. |
PretrainingDataset LAION5B |
LAION-5B open dataset (5 billion pairs). |
PretrainingDataset MetaCLIPCurated |
MetaCLIP metadata-curated dataset. |
PretrainingDataset PMC15M |
PMC-15M biomedical dataset. |
PretrainingDataset WIT400M |
OpenAI’s proprietary WIT (WebImageText) 400M dataset. |
PretrainingDataset WebLI |
WebLI (Web Language-Image) dataset used by Google models. |
int value__ |