Span Identification Adapter

This adapter was trained for span identification in Ukrainian news text.

Usage

from adapters import AutoAdapterModel

# Load base model
model = AutoAdapterModel.from_pretrained("xlm-roberta-base")

# Load this adapter
adapter_name = model.load_adapter("dariast/span-identification-adapter", source="hf")

# Add classification head (if not already present)
model.add_tagging_head(
    adapter_name,
    num_labels=2,
    id2label={0: "NON-SPAN", 1: "SPAN"}
)

# Activate adapter
model.set_active_adapters(adapter_name)

Performance

Best F1 Score: 0.5849
From 5-fold cross-validation
Trained on Ukrainian news span identification task