You can deploy this model and start earning money today!

hadikhamoud/
DRU-RE-one-hot-no-dropout

Name: DRU-RE-one-hot-no-dropout
Availability: InStock

safetensors

Views

No views yet

DRU-RE

Fine-tuned relation extraction model using NeoAraBERT with typed subject/object markers and directional entity-type features.

Base model

U4RASD/NeoAraBERT

Architecture

Input sentence is marked with typed subject/object markers, encoded using NeoAraBERT, then the CLS vector is concatenated with a 42-dimensional subject/object type feature vector.

Final classifier input:

CLS(768) + type_features(42) = 810

Output labels:

41 relation classes.

Important files

model/pytorch_model.bin: custom PyTorch state dict
model/: tokenizer and encoder config
configs/labels.json
configs/entity_types.json
configs/special_marker_tokens.json
configs/run_config.json
artifacts/: metrics and prediction files if generated

Note

This is a custom nn.Module, not a plain AutoModelForSequenceClassification. To reload it, recreate the same model class and load model/pytorch_model.bin.

hadikhamoud/DRU-RE-one-hot-no-dropout