Configuration

BiLSTM (ViroBench classification)

Evidence concern: excluded from comparisons

Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM'

Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)'

BiLSTM: Trained from scratch on the ViroBench training split (non-pretrained baseline).

12 evaluations · 12 results

Overview

An explanatory profile has not yet been reviewed for this record.

Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.

Evaluations and results

12 evaluations · 12 results. Different protocols are not a single leaderboard.

Sorted by Macro F1 (ViroBench ALL viruses, host, G-Split) (higher is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.

Tested configuration	Protocol and dataset	Macro F1 (ViroBench ALL viruses, host, G-Split)	Macro F1 (ViroBench ALL viruses, host, T-Split)	Macro F1 (ViroBench ALL viruses, taxonomy, G-Split)	Macro F1 (ViroBench ALL viruses, taxonomy, T-Split)	Macro F1 (ViroBench DNA viruses, host, G-Split)	Macro F1 (ViroBench DNA viruses, host, T-Split)	Macro F1 (ViroBench DNA viruses, taxonomy, G-Split)	Macro F1 (ViroBench DNA viruses, taxonomy, T-Split)	Macro F1 (ViroBench RNA viruses, host, G-Split)	Macro F1 (ViroBench RNA viruses, host, T-Split)	Macro F1 (ViroBench RNA viruses, taxonomy, G-Split)	Macro F1 (ViroBench RNA viruses, taxonomy, T-Split)
BiLSTM (ViroBench classification)Evaluation	ViroBench ALL viruses host classification, G-Split, Macro-F1, ViroBench ALL viruses, G-Split test set	84.40 (0.98)%	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench ALL viruses host classification, T-Split, Macro-F1, ViroBench ALL viruses, T-Split test set	Not reported	44.69 (1.31)%	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench ALL viruses taxonomy classification, G-Split, Macro-F1, ViroBench ALL viruses, G-Split test set	Not reported	Not reported	66.05 (1.89)%	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench ALL viruses taxonomy classification, T-Split, Macro-F1, ViroBench ALL viruses, T-Split test set	Not reported	Not reported	Not reported	54.67 (2.27)%	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench DNA viruses host classification, G-Split, Macro-F1, ViroBench DNA viruses, G-Split test set	Not reported	Not reported	Not reported	Not reported	62.90 (7.82)%	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench DNA viruses host classification, T-Split, Macro-F1, ViroBench DNA viruses, T-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	56.48 (0.80)%	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench DNA viruses taxonomy classification, G-Split, Macro-F1, ViroBench DNA viruses, G-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	69.67 (3.25)%	Not reported	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench DNA viruses taxonomy classification, T-Split, Macro-F1, ViroBench DNA viruses, T-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	57.79 (2.76)%	Not reported	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench RNA viruses host classification, G-Split, Macro-F1, ViroBench RNA viruses, G-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	81.56 (0.57)%	Not reported	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench RNA viruses host classification, T-Split, Macro-F1, ViroBench RNA viruses, T-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	65.11 (2.04)%	Not reported	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench RNA viruses taxonomy classification, G-Split, Macro-F1, ViroBench RNA viruses, G-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	73.96 (3.79)%	Not reported
BiLSTM (ViroBench classification)Evaluation	ViroBench RNA viruses taxonomy classification, T-Split, Macro-F1, ViroBench RNA viruses, T-Split test set	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	57.43 (1.77)%

All 12 result rows with coverage, uncertainty and sources

Filter evaluations

Tested model or methodProtocolDatasetEvaluation setupMetricEvidence origin

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configuration	Protocol and dataset	Finding	Evidence and details
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench ALL viruses host classification, G-Split, Macro-F1 Dataset: ViroBench ALL viruses, G-Split test set	84.40 (0.98)% Macro F1 (ViroBench ALL viruses, host, G-Split) percent · higher Uncertainty: SD 0.98 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench ALL viruses host classification (G-Split) lit-microbial-20261011-virobench-protocol-all-host-g-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'ALL viruses / Host / G-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench ALL viruses host classification, T-Split, Macro-F1 Dataset: ViroBench ALL viruses, T-Split test set	44.69 (1.31)% Macro F1 (ViroBench ALL viruses, host, T-Split) percent · higher Uncertainty: SD 1.31 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench ALL viruses host classification (T-Split) lit-microbial-20261011-virobench-protocol-all-host-t-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'ALL viruses / Host / T-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench ALL viruses taxonomy classification, G-Split, Macro-F1 Dataset: ViroBench ALL viruses, G-Split test set	66.05 (1.89)% Macro F1 (ViroBench ALL viruses, taxonomy, G-Split) percent · higher Uncertainty: SD 1.89 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench ALL viruses taxonomy classification (G-Split) lit-microbial-20261011-virobench-protocol-all-taxonomy-g-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'ALL viruses / Taxonomy / G-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench ALL viruses taxonomy classification, T-Split, Macro-F1 Dataset: ViroBench ALL viruses, T-Split test set	54.67 (2.27)% Macro F1 (ViroBench ALL viruses, taxonomy, T-Split) percent · higher Uncertainty: SD 2.27 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench ALL viruses taxonomy classification (T-Split) lit-microbial-20261011-virobench-protocol-all-taxonomy-t-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'ALL viruses / Taxonomy / T-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench DNA viruses host classification, G-Split, Macro-F1 Dataset: ViroBench DNA viruses, G-Split test set	62.90 (7.82)% Macro F1 (ViroBench DNA viruses, host, G-Split) percent · higher Uncertainty: SD 7.82 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench DNA viruses host classification (G-Split) lit-microbial-20261011-virobench-protocol-dna-host-g-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'DNA viruses / Host / G-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench DNA viruses host classification, T-Split, Macro-F1 Dataset: ViroBench DNA viruses, T-Split test set	56.48 (0.80)% Macro F1 (ViroBench DNA viruses, host, T-Split) percent · higher Uncertainty: SD 0.80 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench DNA viruses host classification (T-Split) lit-microbial-20261011-virobench-protocol-dna-host-t-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'DNA viruses / Host / T-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench DNA viruses taxonomy classification, G-Split, Macro-F1 Dataset: ViroBench DNA viruses, G-Split test set	69.67 (3.25)% Macro F1 (ViroBench DNA viruses, taxonomy, G-Split) percent · higher Uncertainty: SD 3.25 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench DNA viruses taxonomy classification (G-Split) lit-microbial-20261011-virobench-protocol-dna-taxonomy-g-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'DNA viruses / Taxonomy / G-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench DNA viruses taxonomy classification, T-Split, Macro-F1 Dataset: ViroBench DNA viruses, T-Split test set	57.79 (2.76)% Macro F1 (ViroBench DNA viruses, taxonomy, T-Split) percent · higher Uncertainty: SD 2.76 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench DNA viruses taxonomy classification (T-Split) lit-microbial-20261011-virobench-protocol-dna-taxonomy-t-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'DNA viruses / Taxonomy / T-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench RNA viruses host classification, G-Split, Macro-F1 Dataset: ViroBench RNA viruses, G-Split test set	81.56 (0.57)% Macro F1 (ViroBench RNA viruses, host, G-Split) percent · higher Uncertainty: SD 0.57 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench RNA viruses host classification (G-Split) lit-microbial-20261011-virobench-protocol-rna-host-g-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'RNA viruses / Host / G-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench RNA viruses host classification, T-Split, Macro-F1 Dataset: ViroBench RNA viruses, T-Split test set	65.11 (2.04)% Macro F1 (ViroBench RNA viruses, host, T-Split) percent · higher Uncertainty: SD 2.04 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench RNA viruses host classification (T-Split) lit-microbial-20261011-virobench-protocol-rna-host-t-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'RNA viruses / Host / T-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench RNA viruses taxonomy classification, G-Split, Macro-F1 Dataset: ViroBench RNA viruses, G-Split test set	73.96 (3.79)% Macro F1 (ViroBench RNA viruses, taxonomy, G-Split) percent · higher Uncertainty: SD 3.79 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench RNA viruses taxonomy classification (G-Split) lit-microbial-20261011-virobench-protocol-rna-taxonomy-g-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'RNA viruses / Taxonomy / G-Split' (text-layer line 2534)
Configuration: BiLSTM (ViroBench classification)	Protocol: ViroBench RNA viruses taxonomy classification, T-Split, Macro-F1 Dataset: ViroBench RNA viruses, T-Split test set	57.43 (1.77)% Macro F1 (ViroBench RNA viruses, taxonomy, T-Split) percent · higher Uncertainty: SD 1.77 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 25 prints one Macro-F1 per model for the taxonomy columns. For 143 of the 144 model-by-split taxonomy cells on the ALL-virus set, the printed value equals the mean of the five per-rank values in Table 26 rounded to two decimals. The exception is RNA-FM under the genus-disjoint split: Table 25 prints 48.67 and the mean of the Table 26 ranks (59.96, 65.35, 37.58, 58.78, 43.82) is 53.10. One of the two tables is wrong for that row, so RNA-FM ALL taxonomy G-Split should not be used in a comparison without checking the published code.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'RNA-FM', column 'ALL viruses / Taxonomy / G-Split'; Table 26, row 'RNA-FM' Table 3 lists NT v3 under phage-specific pretraining coverage (P) and Table 9 lists it as non-viral (N). Table 25 groups the NTv3 rows under phage-specific coverage, which is what the configurations record. Pretraining coverage is the paper's main explanatory variable, so NT v3 rows should not be used to support a coverage-versus-performance claim.ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 3 row 'NT v3'; Table 9 row 'NT v3'; Table 25 section 'DNA Foundation Models (Phage-specific Coverage)' Methods, coverage and source BiLSTM on ViroBench RNA viruses taxonomy classification (T-Split) lit-microbial-20261011-virobench-protocol-rna-taxonomy-t-split Aggregation: Not reported ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Table 25, row 'BiLSTM', column 'RNA viruses / Taxonomy / T-Split' (text-layer line 2534)

Source checking is not independent reproduction. Release 2026-10-22-1b7cf027e4f6.

Use this configuration

How it works, versions and access

Strengths, limitations and unresolved questions

Evidence

Source checking verifies the cited claim or transcription. It does not establish independent reproduction.

Evidence table

Inspect claims, sources and review details

Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.

One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.

0 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-10-22-1b7cf027e4f6
Property and statement	Original source and location	Review and provenance

No evidence rows match these filters. Choose another scope or clear the search.

How to read the evidence

Sources and history

Release 2026-10-22-1b7cf027e4f6 · Record review: source checked

1 source record and release history

ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks · Original source · arXiv 2605.25388v1 (submitted 2026-05-25); 42-page PDF; KDD '26 header on every page

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: lit-microbial-20261011-virobench-config-bilstm

areas: microbes-communities; dna-genomes
contexts: research
method types: supervised_machine_learning
reported name: BiLSTM
foundation model eligible: false
parameters: Trained from scratch on the ViroBench training split (non-pretrained baseline)
source locator: Table 25 row 'BiLSTM' (section 'Baseline'); Table 9
access reuse: Not stated in the source; see the model's own release
missing metadata: version: reason: unreported; note: The printed row label is the only checkpoint identifier; Table 9 lists evaluated sub-modes
provenance note: BiLSTM as evaluated in ViroBench viral taxonomy and host classification.

Related records

Suggest a correction