Phenotype Mining on BiolarkGSC+
0.603F1 ScoreFastHPOCR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FastHPOCRFT=✗2025.07 | 0.603 | 0.721 | 0.518 | |
| FastHPOCRApproach=Baselines2025.07 | 0.603 | — | — | |
| RDMA (Mistral 24B)FT=✗2025.07 | 0.559 | 0.565 | 0.553 | |
| Mistral 24BApproach=RDMA2025.07 | 0.559 | — | — | |
| GPT-5Approach=RDMA2025.07 | 0.545 | — | — | |
| GPT-5Approach=RAG2025.07 | 0.525 | — | — | |
| BioBERTFT=✓2025.07 | 0.498 | 0.635 | 0.409 | |
| BioBERT-MRCApproach=Baselines2025.07 | 0.498 | — | — | |
| i2b2 Clinical BERTFT=✓2025.07 | 0.491 | 0.599 | 0.417 | |
| i2b2Approach=Baselines2025.07 | 0.491 | — | — | |
| Llama-3 70BApproach=RDMA2025.07 | 0.484 | — | — | |
| RAG (RD / HPO) (Llama 3.3 70B)FT=✗2025.07 | 0.475 | 0.41 | 0.563 | |
| Llama-3 70BApproach=RAG2025.07 | 0.474 | — | — | |
| Qwen3 32BApproach=RDMA2025.07 | 0.473 | — | — | |
| BioClinicalBERTFT=✓2025.07 | 0.441 | 0.459 | 0.424 | |
| BioClinicalBERTApproach=Baselines2025.07 | 0.441 | — | — | |
| PhenoGPTFT=✓2025.07 | 0.427 | 0.676 | 0.312 | |
| PhenoGPTApproach=Baselines2025.07 | 0.427 | — | — | |
| Qwen3 32BApproach=RAG2025.07 | 0.401 | — | — | |
| Mistral 24BApproach=RAG2025.07 | 0.398 | — | — | |
| Llama-3 8BApproach=RAG2025.07 | 0.345 | — | — | |
| Llama-3 8BApproach=RDMA2025.07 | 0.331 | — | — | |
| Dictionary MatchFT=✗2025.07 | 0.326 | 0.682 | 0.214 | |
| DictionaryApproach=Baselines2025.07 | 0.326 | — | — | |
| Zero-Shot (Llama 3.3 70B)FT=✗2025.07 | 0 | 0 | 0 | |
| GPT-5Approach=Zeroshot2025.07 | 0 | — | — | |
| Llama-3 70BApproach=Zeroshot2025.07 | 0 | — | — | |
| Llama-3 8BApproach=Zeroshot2025.07 | 0 | — | — | |
| Mistral 24BApproach=Zeroshot2025.07 | 0 | — | — | |
| Qwen3 32BApproach=Zeroshot2025.07 | 0 | — | — |