Named Entity Recognition on OntoNotes (Micro-F1)
89Micro-F1GLiNER
Evaluation results
| Method | Links | |
|---|---|---|
| GLiNERBackbone=deBERTa-v3, #Trained Params=209M2025.10 | 89 | |
| ToMMeR + span embed (ℓ11)Backbone=RoBERTa-base, #Trained Params=7.4M, Probed layer=ℓ52025.10 | 85.4 | |
| ToMMeR + span embed (ℓ11)Backbone=BERT-base, #Trained Params=7.4M, Probed layer=ℓ32025.10 | 82.1 | |
| ToMMeR + span embed (ℓ11)Backbone=LLaMA-3.2-3B, #Trained Params=7.7M, Probed layer=ℓ52025.10 | 81.7 | |
| ToMMeR + span embed (ℓ11)Backbone=LLaMA-3.2-1B, #Trained Params=7.6M, Probed layer=ℓ62025.10 | 80.4 | |
| ToMMeR + span embed (ℓ11)Backbone=LLaMA-3.1-8B, #Trained Params=7.9M, Probed layer=ℓ52025.10 | 80 | |
| EMBERBackbone=GPT2-xl (0–47), #Trained Params=11.5M2025.10 | 79.3 |