Span Prediction on Drug-ACE (test)
51.43Hard F1Gemma3-4BRCLoRA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Gemma3-4BRCLoRATrainable Params=15.46 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 51.43 | 59.74 | |
| Qwen3-4BRCLoRATrainable Params=17.11 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 51.18 | 60.62 | |
| Gemma2-9BRCLoRATrainable Params=28.39 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 49.89 | 59.57 | |
| Qwen3-4BLoRATrainable Params=16.52 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 49.37 | 59.84 | |
| Gemma3-4BLoRATrainable Params=14.90 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 49.17 | 56.53 | |
| MedGemma-4BRCLoRATrainable Params=15.46 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 48.63 | 57.4 | |
| Gemma2-9BLoRATrainable Params=27.01 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 48.03 | 59.09 | |
| Qwen2.5-7BRCLoRATrainable Params=20.42 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 46.82 | 56.99 | |
| MedGemma-4BLoRATrainable Params=14.90 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 46.62 | 55.35 | |
| Qwen2.5-7BLoRATrainable Params=20.19 M, Evaluation Protocol=LoRA Fine-tuning2026.06 | 46.53 | 56.12 | |
| BiomedBERT-baseTrainable Params=108.26 M, Evaluation Protocol=SpanMarker2026.06 | 39.35 | 53.23 | |
| RoBERTa-largeTrainable Params=355.39 M, Evaluation Protocol=SpanMarker2026.06 | 37.81 | 52.16 | |
| RoBERTa-baseTrainable Params=124.67 M, Evaluation Protocol=SpanMarker2026.06 | 34.86 | 48 | |
| BioBERT-base-casedTrainable Params=108.33 M, Evaluation Protocol=SpanMarker2026.06 | 33.57 | 45.49 | |
| BERT-base-uncasedTrainable Params=109.51 M, Evaluation Protocol=SpanMarker2026.06 | 33.07 | 45.2 | |
| Bio_ClinicalBERTTrainable Params=108.33 M, Evaluation Protocol=SpanMarker2026.06 | 32.96 | 46.4 | |
| BERT-base-casedTrainable Params=108.33 M, Evaluation Protocol=SpanMarker2026.06 | 30.73 | 45.39 | |
| Llama3.3-70BEvaluation Protocol=2-Shot Prompting2026.06 | 23.75 | 32.37 | |
| Qwen2.5-72BEvaluation Protocol=2-Shot Prompting2026.06 | 23.04 | 31.35 | |
| DeepSeek-R1-70BEvaluation Protocol=2-Shot Prompting2026.06 | 13.91 | 35.31 |