| task | ours | baseline | margin | |
|---|---|---|---|---|
| compound class: BGC-1, 2048-base windows | 0.7406 | 0.4625 | nearest known neighbour | 0.2781 |
| compound class: BGC-1, 2048-base windows | 0.7406 | 0.5031 | EH14 X on the same task | 0.2375 |
| compound class: BGC-1, 512-base windows | 0.6937 | 0.4625 | nearest known neighbour | 0.2312 |
| name | BGC-1 |
| parameters | 321,017,986 |
| window | 2,048 |
| corpus | bgcgroot |
| tokens read | 5,520,000,000 |
| points[0] › tokens_per_rang | 40,000,000 |
| points[0] › auc | 0.4791 |
| points[0] › loss | 1.2603 |
| points[0] › skipped | 0 |
| points[0] › minutes | 73 |
| points[1] › tokens_per_rang | 80,000,000 |
| points[1] › auc | 0.4867 |
| points[1] › loss | 1.2624 |
| points[1] › skipped | 0 |
| points[1] › minutes | 145 |
| points[2] › tokens_per_rang | 160,000,000 |
| points[2] › auc | 0.4946 |
| points[2] › loss | 1.2181 |
| points[2] › skipped | 0 |
| points[2] › minutes | 291 |
| points[3] › tokens_per_rang | 240,000,000 |
| points[3] › auc | 0.4932 |
| points[3] › loss | 1.1907 |
| points[3] › skipped | 0 |
| points[3] › minutes | 436 |
| points[4] › tokens_per_rang | 320,000,000 |
| points[4] › auc | 0.4805 |
| points[4] › loss | 1.1751 |
| points[4] › skipped | 0 |
| points[4] › minutes | 581 |
| points[5] › tokens_per_rang | 410,000,000 |
| points[5] › auc | 0.4866 |
| points[5] › loss | 1.1445 |
| points[5] › skipped | 0 |
| points[5] › minutes | 726 |
| points[6] › tokens_per_rang | 490,000,000 |
| points[6] › auc | 0.4895 |
| points[6] › loss | 1.114 |
| points[6] › skipped | 0 |
| points[6] › minutes | 871 |
| points[7] › tokens_per_rang | 570,000,000 |
| points[7] › auc | 0.4907 |
| points[7] › loss | 1.1126 |
| points[7] › skipped | 0 |
| points[7] › minutes | 1,016 |
| points[8] › tokens_per_rang | 690,000,000 |
| points[8] › auc | 0.4885 |
| points[8] › loss | 1.0876 |
| points[8] › skipped | 0 |
| points[8] › minutes | 1,233 |
| final loss | 1.0876 |
| perplexity | 2.967 |
| best auc | 0.4946 |
| tokens | 5,520,000,000 |
| parameters of the log | 320,753,793 |
| clusters | 1,604 |
| families | 1,534 |
| train | 1,284 |
| test | 320 |
| measurement › largest class guessing › correct | 0.2281 |
| measurement › largest class guessing › interval | 0.1812 – 0.2751 |
| measurement › 4-mers + linear layer › correct | 0.4594 |
| measurement › 4-mers + linear layer › interval | 0.4093 – 0.5125 |
| measurement › EH14 X, 512-base windows › correct | 0.5031 |
| measurement › EH14 X, 512-base windows › interval | 0.4531 – 0.5563 |
| measurement › BGC-1, 512-base windows › correct | 0.6937 |
| measurement › BGC-1, 512-base windows › interval | 0.6468 – 0.7438 |
| measurement › BGC-1, 2048-base windows › correct | 0.7406 |
| measurement › BGC-1, 2048-base windows › interval | 0.6906 – 0.7875 |
| measured with | EH14 X, 6.7 M (previous version) – BGC-1 |
| prefix | 512 |
| written | 1,536 |
| families | 1,534 |
| per class | 40 |
| stops floor random | 11.74 |
| classes › NRPS › n | 40 |
| classes › NRPS › model | 0.325 |
| classes › NRPS › markov | 0.125 |
| classes › NRPS › real held-out | 0.375 |
| classes › NRPS › stops model | 0.65 |
| classes › NRPS › orf_model | 1,437 |
| classes › NRPS › stops markov | 5.22 |
| classes › NRPS › orf_markov | 570 |
| classes › NRPS › stops real | 0 |
| classes › NRPS › orf real | 1,533 |
| classes › NRPS › distance model | 0.65 |
| classes › NRPS › distance markov | 5.22 |
| classes › PKS › n | 40 |
| classes › PKS › model | 0.15 |
| classes › PKS › markov | 0.225 |
| classes › PKS › real held-out | 0.375 |
| classes › PKS › stops model | 1.3 |
| classes › PKS › orf_model | 1,156 |
| classes › PKS › stops markov | 5.21 |
| classes › PKS › orf_markov | 550 |
| classes › PKS › stops real | 0.65 |
| classes › PKS › orf real | 1,353 |
| classes › PKS › distance model | 0.65 |
| classes › PKS › distance markov | 4.56 |
| classes › other › n | 40 |
| classes › other › model | 0.125 |
| classes › other › markov | 0.2 |
| classes › other › real held-out | 0.325 |
| classes › other › stops model | 0 |
| classes › other › orf_model | 1,533 |
| classes › other › stops markov | 5.54 |
| classes › other › orf_markov | 538 |
| classes › other › stops real | 1.3 |
| classes › other › orf real | 1,225 |
| classes › other › distance model | 1.3 |
| classes › other › distance markov | 4.24 |
| classes › ribosomal › n | 40 |
| classes › ribosomal › model | 0.125 |
| classes › ribosomal › markov | 0.025 |
| classes › ribosomal › real held-out | 0.35 |
| classes › ribosomal › stops model | 0.65 |
| classes › ribosomal › orf_model | 1,341 |
| classes › ribosomal › stops markov | 9.78 |
| classes › ribosomal › orf_markov | 418 |
| classes › ribosomal › stops real | 1.3 |
| classes › ribosomal › orf real | 1,273 |
| classes › ribosomal › distance model | 0.65 |
| classes › ribosomal › distance markov | 8.48 |
| classes › saccharide › n | 33 |
| classes › saccharide › model | 0.1818 |
| classes › saccharide › markov | 0.1212 |
| classes › saccharide › real held-out | 0.2424 |
| classes › saccharide › stops model | 0.65 |
| classes › saccharide › orf_model | 1,389 |
| classes › saccharide › stops markov | 8.46 |
| classes › saccharide › orf_markov | 417 |
| classes › saccharide › stops real | 2.6 |
| classes › saccharide › orf real | 1,017 |
| classes › saccharide › distance model | 1.95 |
| classes › saccharide › distance markov | 5.86 |
| classes › terpene › n | 19 |
| classes › terpene › model | 0.2105 |
| classes › terpene › markov | 0.2105 |
| classes › terpene › real held-out | 0.2105 |
| classes › terpene › stops model | 2.61 |
| classes › terpene › orf_model | 975 |
| classes › terpene › stops markov | 8.46 |
| classes › terpene › orf_markov | 441 |
| classes › terpene › stops real | 7.18 |
| classes › terpene › orf real | 759 |
| classes › terpene › distance model | 4.57 |
| classes › terpene › distance markov | 1.28 |
| model mean | 0.1862 |
| markov mean | 0.1511 |
| real mean | 0.313 |
| orf model mean | 1,305.1667 |
| orf markov mean | 489 |
| orf real mean | 1,193.3333 |
| stops model mean | 0.9767 |
| stops markov mean | 7.1117 |
| stops real mean | 2.1717 |
| distance model mean | 1.6283 |
| distance markov mean | 4.94 |
| difference | 0.0351 |
| baseline 4-mers 512 | 0.8227 |
| eh14x_fp32 | 0.7063 |
| draws | 2,000 |
| vol › 6.4M | 0.6288 |
| vol › 50M | 0.6542 |
| vol › difference | 0.0254 |
| vol › interval | 0.0026 – 0.0494 |
| real › 6.4M | 0.642 |
| real › 50M | 0.6931 |
| real › difference | 0.0511 |
| real › interval | 0.0371 – 0.0656 |
| length | 200 |
| groups › dinoflagellata › taxonomic placement › real held-out | 0.9667 |
| groups › dinoflagellata › taxonomic placement › markov | 0.6167 |
| groups › dinoflagellata › taxonomic placement › model | 0.1 |
| groups › dinoflagellata › similarity › real held-out | 0.8955 |
| groups › dinoflagellata › similarity › markov | 0.6347 |
| groups › dinoflagellata › similarity › model | 0.5789 |
| groups › dinoflagellata › model lands at › fungi | 21 |
| groups › dinoflagellata › model lands at › apicomplexa | 10 |
| groups › dinoflagellata › model lands at › dinoflagellata | 6 |
| groups › dinoflagellata › model lands at › gyrista | 5 |
| groups › dinoflagellata › model lands at › chlorophyta | 5 |
| groups › dinoflagellata › model lands at › discoba | 4 |
| groups › dinoflagellata › model lands at › ciliophora | 4 |
| groups › dinoflagellata › model lands at › metazoa | 4 |
| groups › dinoflagellata › model lands at › rhodophyta | 1 |
| groups › dinoflagellata › too separate › markov › auc | 0.9983 |
| groups › dinoflagellata › too separate › markov › interval | 0.995 – 1 |
| groups › dinoflagellata › too separate › model › auc | 1 |
| groups › dinoflagellata › too separate › model › interval | 1 – 1 |
| groups › dinoflagellata › model failed | 0 |
| groups › ciliophora › taxonomic placement › real held-out | 0.9833 |
| groups › ciliophora › taxonomic placement › markov | 0.5167 |
| groups › ciliophora › taxonomic placement › model | 0.1333 |
| groups › ciliophora › similarity › real held-out | 0.9 |
| groups › ciliophora › similarity › markov | 0.6845 |
| groups › ciliophora › similarity › model | 0.6175 |
| groups › ciliophora › model lands at › fungi | 19 |
| groups › ciliophora › model lands at › apicomplexa | 17 |
| groups › ciliophora › model lands at › ciliophora | 8 |
| groups › ciliophora › model lands at › gyrista | 7 |
| groups › ciliophora › model lands at › rhodophyta | 2 |
| groups › ciliophora › model lands at › discoba | 2 |
| groups › ciliophora › model lands at › dinoflagellata | 2 |
| groups › ciliophora › model lands at › metazoa | 2 |
| groups › ciliophora › model lands at › chlorophyta | 1 |
| groups › ciliophora › too separate › markov › auc | 0.9961 |
| groups › ciliophora › too separate › markov › interval | 0.9872 – 1 |
| groups › ciliophora › too separate › model › auc | 1 |
| groups › ciliophora › too separate › model › interval | 1 – 1 |
| groups › ciliophora › model failed | 0 |
| groups › metazoa › taxonomic placement › real held-out | 0.9667 |
| groups › metazoa › taxonomic placement › markov | 0.7167 |
| groups › metazoa › taxonomic placement › model | 0.0667 |
| groups › metazoa › similarity › real held-out | 0.8384 |
| groups › metazoa › similarity › markov | 0.6636 |
| groups › metazoa › similarity › model | 0.583 |
| groups › metazoa › model lands at › fungi | 21 |
| groups › metazoa › model lands at › apicomplexa | 15 |
| groups › metazoa › model lands at › ciliophora | 6 |
| groups › metazoa › model lands at › rhodophyta | 5 |
| groups › metazoa › model lands at › metazoa | 4 |
| groups › metazoa › model lands at › discoba | 3 |
| groups › metazoa › model lands at › dinoflagellata | 2 |
| groups › metazoa › model lands at › gyrista | 2 |
| groups › metazoa › model lands at › chlorophyta | 2 |
| groups › metazoa › too separate › markov › auc | 0.9381 |
| groups › metazoa › too separate › markov › interval | 0.8902 – 0.9762 |
| groups › metazoa › too separate › model › auc | 0.9956 |
| groups › metazoa › too separate › model › interval | 0.9833 – 1 |
| groups › metazoa › model failed | 0 |
| measured with | no model |
| prefix | 512 |
| written | 1,536 |
| families | 1,534 |
| per class | 40 |
| classes › NRPS › n | 40 |
| classes › NRPS › model | — |
| classes › NRPS › markov | 0.125 |
| classes › NRPS › real held-out | 0.375 |
| classes › PKS › n | 40 |
| classes › PKS › model | — |
| classes › PKS › markov | 0.225 |
| classes › PKS › real held-out | 0.375 |
| classes › other › n | 40 |
| classes › other › model | — |
| classes › other › markov | 0.2 |
| classes › other › real held-out | 0.325 |
| classes › ribosomal › n | 40 |
| classes › ribosomal › model | — |
| classes › ribosomal › markov | 0.025 |
| classes › ribosomal › real held-out | 0.35 |
| classes › saccharide › n | 33 |
| classes › saccharide › model | — |
| classes › saccharide › markov | 0.1212 |
| classes › saccharide › real held-out | 0.2424 |
| classes › terpene › n | 19 |
| classes › terpene › model | — |
| classes › terpene › markov | 0.2105 |
| classes › terpene › real held-out | 0.2105 |
| model mean | — |
| markov mean | 0.1511 |
| real mean | 0.313 |
| size | 7,171,796,732 |
| rules | 12,824,758 |
| on | 02-10-2026 13:18 |
| measured with | no model |
| bgc-1 › gpu hours | 180.85 |
| bgc-1 › gpu minutes | 10,851 |
| bgc-1 › cpu hours | 0 |
| eh14x › gpu hours | 0 |
| eh14x › gpu minutes | 0 |
| eh14x › cpu hours | 22.2 |
| eh14x › cpu minutes | 1,331 |