Protein Fold Classification on TEDBench (test)
88.92AccuracyESM2-15B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ESM2-15Bsize=15B, Evaluation Protocol=pretrained + linear probing2026.05 | 88.92 | — | 76.27 | |
| ESM2-3Bsize=3B, Evaluation Protocol=pretrained + linear probing2026.05 | 88.82 | — | 75.75 | |
| SaProt-650Msize=650M, Evaluation Protocol=pretrained + linear probing2026.05 | 87.31 | — | 70.79 | |
| ESM2-650Msize=650M, Evaluation Protocol=pretrained + linear probing2026.05 | 87.15 | — | 70.03 | |
| MiAE-Lsize=339M, Evaluation Protocol=pretrained + linear probing2026.05 | 87.06 | — | 70.44 | |
| MiAE-B+seqsize=102M, Evaluation Protocol=pretrained + linear probing2026.05 | 86.2 | — | 68.88 | |
| SaProt-35Msize=35M, Evaluation Protocol=pretrained + linear probing2026.05 | 85.49 | — | 67.08 | |
| MiAE-Bsize=102M, Evaluation Protocol=pretrained + linear probing2026.05 | 85.31 | — | 66.18 | |
| ESM2-150Msize=150M, Evaluation Protocol=pretrained + linear probing2026.05 | 84.71 | — | 63.36 | |
| MiAE-Ssize=29M, Evaluation Protocol=pretrained + linear probing2026.05 | 80.35 | — | 59.03 | |
| ESM2-35Msize=35M, Evaluation Protocol=pretrained + linear probing2026.05 | 79.48 | — | 52.66 | |
| ESM2-3Bsize=3B, Evaluation Protocol=pretrained + linear probing2026.05 | 76.44 | — | 69.08 | |
| ESM2-15Bsize=15B, Evaluation Protocol=pretrained + linear probing2026.05 | 76.32 | — | 70.85 | |
| MiAE-Lsize=339M, Evaluation Protocol=pretrained + linear probing2026.05 | 75.42 | — | 63.5 | |
| SaProt-650Msize=650M, Evaluation Protocol=pretrained + linear probing2026.05 | 75.23 | — | 66.55 | |
| ESM2-650Msize=650M, Evaluation Protocol=pretrained + linear probing2026.05 | 74.65 | — | 62.32 | |
| MiAE-B+seqsize=102M, Evaluation Protocol=pretrained + linear probing2026.05 | 74.08 | — | 62.14 | |
| SaProt-35Msize=35M, Evaluation Protocol=pretrained + linear probing2026.05 | 73.06 | — | 58.82 | |
| MiAE-Bsize=102M, Evaluation Protocol=pretrained + linear probing2026.05 | 72.94 | — | 58.52 | |
| ESM2-150Msize=150M, Evaluation Protocol=pretrained + linear probing2026.05 | 70.81 | — | 54.39 | |
| MiAE-Ssize=29M, Evaluation Protocol=pretrained + linear probing2026.05 | 67.9 | — | 49.43 | |
| ESM2-35Msize=35M, Evaluation Protocol=pretrained + linear probing2026.05 | 65.71 | — | 41.91 | |
| PMPNNsize=1.6M, Evaluation Protocol=pretrained + linear probing2026.05 | 59.88 | — | 38.92 | |
| MIFsize=3.4M, Evaluation Protocol=pretrained + linear probing2026.05 | 56.49 | — | 44.38 | |
| PMPNNsize=1.6M, Evaluation Protocol=pretrained + linear probing2026.05 | 54.25 | — | 41.43 | |
| MIFsize=3.4M, Evaluation Protocol=pretrained + linear probing2026.05 | 52.02 | — | 34.36 | |
| SaProt-650Msize=650M, training protocol=pretrained + finetuned2026.05 | 0.8031 | 0.7348 | — | |
| MiAE-B+seqsize=102M, training protocol=pretrained + finetuned2026.05 | 0.8031 | 0.7456 | — | |
| MiAE-Lsize=339M, training protocol=pretrained + finetuned2026.05 | 0.801 | 0.7347 | — | |
| MiAE-Bsize=102M, training protocol=pretrained + finetuned2026.05 | 0.799 | 0.7371 | — | |
| SaProt-35Msize=35M, training protocol=pretrained + finetuned2026.05 | 0.7946 | 0.7175 | — | |
| MiAE-Ssize=29M, training protocol=pretrained + finetuned2026.05 | 0.7915 | 0.7228 | — | |
| MiAE-Bsize=102M, training protocol=supervised from scratch2026.05 | 0.7836 | 0.716 | — | |
| MiAE-Lsize=339M, training protocol=supervised from scratch2026.05 | 0.7834 | 0.7095 | — | |
| MiAE-B+seqsize=102M, training protocol=supervised from scratch2026.05 | 0.7823 | 0.7164 | — | |
| MiAE-Ssize=29M, training protocol=supervised from scratch2026.05 | 0.7805 | 0.7003 | — | |
| ESM2-650Msize=650M, training protocol=pretrained + finetuned2026.05 | 0.7663 | 0.6619 | — | |
| GotenNetsize=1.9M, training protocol=supervised from scratch2026.05 | 0.7333 | 0.6402 | — | |
| ESM2-150Msize=150M, training protocol=pretrained + finetuned2026.05 | 0.7326 | 0.5707 | — | |
| E3NNsize=1.9M, training protocol=supervised from scratch2026.05 | 0.7187 | 0.5763 | — | |
| ESM2-35Msize=35M, training protocol=pretrained + finetuned2026.05 | 0.6814 | 0.4647 | — | |
| MACEsize=1.5M, training protocol=supervised from scratch2026.05 | 0.6689 | 0.5058 | — |