Nucleotide Transformer v2 and Evo: Zero-shot benchmarking of RNA language models in structural, functional, and evolutionary learning reverses the order
- Low severity
- Not yet reviewed
- Ordering reversal
- Different tasks across studies
Zero-shot benchmarking of RNA language models in structural, functional, and evolutionary learning ranks Evo above Nucleotide Transformer v2 in 6 of 6 comparisons; other studies rank Nucleotide Transformer v2 above Evo in 42 of 45.
An automatic lead, not a finding of error. The results below are shown as published, everywhere on the site, until a reviewer decides otherwise.
Which studies agree
GENEB leaderboard ranks Nucleotide Transformer v2 above Evo in 42 of its 45 comparisons and Evo higher in 3. Zero-shot benchmarking of RNA language… ranks Evo above Nucleotide Transformer v2 in all 6 of its comparisons.
- Rank Nucleotide Transformer v2 higher (1): GENEB leaderboard
- Rank Evo higher (1): Zero-shot benchmarking of RNA language… (disagrees)
- Mixed (0): none
Each paper sits where at least two thirds of its comparisons point; Mixed means neither model reaches two thirds. The larger, outlined dot is the paper that disagrees.
| Paper | Role | Ranks Nucleotide Transformer v2 higher | Ranks Evo higher | Ties |
|---|---|---|---|---|
| Zero-shot benchmarking of RNA language models in structural, functional, and evolutionary learning and 1 more record · DOI | Disagrees | 0 | 6 | 0 |
| GENEB leaderboard (leaderboard.json, 13 task-type aggregates) · Repository | Agrees | 42 | 3 | 0 |
Comparisons in Zero-shot benchmarking of RNA language… (6)
Comparisons in GENEB leaderboard (45)
| Protocol | Dataset | Metric | Nucleotide Transformer v2 | Evo | Higher |
|---|---|---|---|---|---|
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Chromatin Acc.'), unitless | 0.481 | 0.289 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Coding/NC'), unitless | 0.587 | 0.539 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'DNA Methyl.'), unitless | 0.0804 | 0.0458 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Enhancers'), unitless | 0.3 | 0.288 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Histone Mod.'), unitless | 0.261 | 0.258 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Mouse Enh.'), unitless | 0.171 | 0.12 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Promoters'), unitless | 0.545 | 0.436 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Regulatory'), unitless | 0.182 | 0.125 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Species Clf.'), unitless | 0.564 | 0.207 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Splice Sites'), unitless | 0.129 | 0.0657 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'TF Binding'), unitless | 0.242 | 0.0386 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'Virus/Phage'), unitless | 0.147 | 0.0941 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; category 'lncRNA'), unitless | 0.0528 | 0.0189 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; macro average: unweighted mean of the 13 category MCCs), unitless | 0.286 | 0.194 | Nucleotide Transformer v2 |
| GENEB leaderboard, 10-shot, MCC aggregates | GENEB 100 tasks | MCC (10-shot; micro average: mean over all 100 tasks), unitless | 0.293 | 0.232 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Chromatin Acc.'), unitless | 0.186 | 0.14 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Coding/NC'), unitless | 0.284 | 0.293 | Evo |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'DNA Methyl.'), unitless | 0.0223 | 0.0208 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Enhancers'), unitless | 0.159 | 0.165 | Evo |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Histone Mod.'), unitless | 0.115 | 0.108 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Mouse Enh.'), unitless | 0.0424 | 0.0568 | Evo |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Promoters'), unitless | 0.36 | 0.279 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Regulatory'), unitless | 0.113 | 0.0788 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Species Clf.'), unitless | 0.304 | 0.103 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Splice Sites'), unitless | 0.0704 | 0.0643 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'TF Binding'), unitless | 0.0992 | 0.0084 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'Virus/Phage'), unitless | 0.0356 | -0.00343 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; category 'lncRNA'), unitless | 0.037 | 0.00964 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; macro average: unweighted mean of the 13 category MCCs), unitless | 0.136 | 0.102 | Nucleotide Transformer v2 |
| GENEB leaderboard, 1-shot, MCC aggregates | GENEB 100 tasks | MCC (1-shot; micro average: mean over all 100 tasks), unitless | 0.158 | 0.126 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Chromatin Acc.'), unitless | 0.652 | 0.284 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Coding/NC'), unitless | 0.812 | 0.719 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'DNA Methyl.'), unitless | 0.339 | 0.0732 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Enhancers'), unitless | 0.469 | 0.37 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Histone Mod.'), unitless | 0.473 | 0.397 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Mouse Enh.'), unitless | 0.492 | 0.287 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Promoters'), unitless | 0.722 | 0.537 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Regulatory'), unitless | 0.457 | 0.269 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Species Clf.'), unitless | 0.747 | 0.285 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Splice Sites'), unitless | 0.47 | 0.16 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'TF Binding'), unitless | 0.55 | 0.173 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'Virus/Phage'), unitless | 0.482 | 0.205 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; category 'lncRNA'), unitless | 0.337 | 0.119 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; macro average: unweighted mean of the 13 category MCCs), unitless | 0.537 | 0.298 | Nucleotide Transformer v2 |
| GENEB leaderboard, full training data, MCC aggregates | GENEB 100 tasks | MCC (full training data; micro average: mean over all 100 tasks), unitless | 0.524 | 0.342 | Nucleotide Transformer v2 |
What disagrees
- Other studies
- rank Nucleotide Transformer v2 above Evo in 42 of 45 comparisons
- This study
- ranks Evo above Nucleotide Transformer v2 in 6 of 6 comparisons
- Compared
- Nucleotide Transformer v2, Evo
- Study that disagrees
- Zero-shot benchmarking of RNA language models in structural, functional, and evolutionary learning, Zero-shot benchmarking of RNA language models: Supplementary Information
- Other studies
- GENEB leaderboard (leaderboard.json, 13 task-type aggregates)
- Rule
- Ordering reversal; different tasks across studies
- Model identity
- Matched across studies partly through links not yet reviewed
The compared numbers
Each value links to its result record and source. Compare values within one protocol; the rule compares rankings, not raw values across protocols.
In the other studies (90 results)
Follow-up
Current state: Not yet reviewed.
No one has reviewed this finding yet.
How to help resolve this
If you know why these studies disagree, for example a different population, metric or preprocessing step, or you have a corrected value, tell us through the contribution form. A reviewer checks every submission against its source before anything changes, and the step is added to the follow-up above.
Finding discrepancy-a1e76daf59cd0666 · All contested results · How findings are made and followed up