Configuration: CMLM (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +52.1 [36.8, 69.1] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is CMLM)unitless · higher
Uncertainty: 95% CI 36.8 to 69.1. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus CMLM on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'CMLM', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: DPLM-2 3B (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +87.6 [66.8, 110.1] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is DPLM-2 3B)unitless · higher
Uncertainty: 95% CI 66.8 to 110.1. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus DPLM-2 3B on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'DPLM-2 3B', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: LigandMPNN (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +20.1 [8.4, 33.0] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is LigandMPNN)unitless · higher
Uncertainty: 95% CI 8.4 to 33.0. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus LigandMPNN on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'LigandMPNN', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: LM-Design 650M (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +49.6 [34.4, 65.7] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is LM-Design 650M)unitless · higher
Uncertainty: 95% CI 34.4 to 65.7. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus LM-Design 650M on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'LM-Design 650M', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: PiFold (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +49.5 [34.6, 66.6] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is PiFold)unitless · higher
Uncertainty: 95% CI 34.6 to 66.6. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus PiFold on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'PiFold', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: ProteinMPNN grouped (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +13.6 [3.5, 25.1] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is ProteinMPNN grouped)unitless · higher
Uncertainty: 95% CI 3.5 to 25.1. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus ProteinMPNN grouped on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'ProteinMPNN grouped', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: ProteinMPNN (low-noise, grouped) (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +16.0 [5.8, 27.4] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is ProteinMPNN (low-noise, grouped))unitless · higher
Uncertainty: 95% CI 5.8 to 27.4. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus ProteinMPNN (low-noise, grouped) on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'ProteinMPNN (low-noise, grouped)', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: ProteinMPNN (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +12.3 [3.8, 21.6] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is ProteinMPNN)unitless · higher
Uncertainty: 95% CI 3.8 to 21.6. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus ProteinMPNN on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'ProteinMPNN', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: ProteinZero V2 (ESM1) (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +11.2 [3.5, 19.3] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is ProteinZero V2 (ESM1))unitless · higher
Uncertainty: 95% CI 3.5 to 19.3. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus ProteinZero V2 (ESM1) on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'ProteinZero V2 (ESM1)', column 'RecentPDB, Boltz2' (page 16 of 21) Configuration: ProteinZero V2 (ESM2) (HyInv report) Protocol: Paired scTM difference from HyInv under Boltz2 (HyInv report) Dataset: RecentPDB (121 chains from 111 recent PDB entries) +10.4 [1.7, 20.0] Sctm difference (HyInv default minus the baseline, paired by chain on RecentPDB with shared folding randomness, printed in units of 10^-3 scTM with a 95% percentile interval; Boltz2 as the structure predictor; the reference system is the HyInv default setting; the baseline is ProteinZero V2 (ESM2))unitless · higher
Uncertainty: 95% CI 1.7 to 20.0. Printed in units of 10^-3 scTM, as the point estimate is. The interval excludes zero.
Coverage: Not reported scored / Not reported eligible
Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Appendix C states 'We do not claim zero sequence overlap between the training data and the evaluation panels', and Appendix G's homology screen is a coverage rule under which partial hits remain (Table 10: 28 of 74 de novo backbones and 7 of 12 pocket backbones have partial hits, and all 4 named enzymes are homologous to training data by design). Results on these panels are therefore not clean held-out measurements and must not be compared with results from sources that do establish no overlap.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix C 'Data and pool', final bullet; Appendix G 'Homology of the panels' and Table 10
The ten baselines are single published settings at one temperature each, so their design-set diversity differs from HyInv's: on CATH4.3 test HyInv samples at diversity 0.331 against 0.093 for PiFold, 0.104 for CMLM, 0.144 for LM-Design 650M and 0.000 for DPLM-2 3B, which decodes by argmax. Appendix D.2 states that these four differences are not at matched diversity, and Appendix A states that even for the closest baselines the comparisons are 'close to, but not exactly at, equal diversity', the largest offset of 0.016 being worth at most 0.0011 scTM at the development slope. The scTM and scTM difference results of the protocols named here must not be read as a matched-diversity ranking. The results read at a stated diversity level (Tables 8 and 9 and Appendix D.3) are not covered by this entry, because they are the ones the report reads on a fitted curve at a fixed diversity; nor are the recovery, diversity, pLDDT, scRMSD and lDDT results, whose protocols carry the same single-setting caveat in their own limitations and whose metric qualifiers say the value is at each method's own setting.HyInv Technical Report (Hyra Team, 2026-10-14) · Appendix D.1 and D.2; Table 7; Figure 14 diversity block; Section 5.1 'Baselines' and 'HyInv'
The inference-time results were measured with the supervised-stage weights and a rewrite temperature of 0.23, not with the released checkpoint and its default temperatures, and the quality read-outs of the two numerical paths used the RL checkpoint at update 288, not update 640. Speed and quality in this report are therefore measured on different weights, and the 57x figure compares HyInv's optimised BF16 path with ProteinMPNN's released implementation rather than with an equally optimised baseline.HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1 footnote; Appendix F 'Timing protocol' and 'Baseline timings'
Methods, coverage and source HyInv minus ProteinZero V2 (ESM2) on RecentPDB Boltz2 (paired scTM difference, HyInv report)
hyinv-20261106-protocol-sctm-difference-boltz2
Aggregation: Not reported
HyInv Technical Report (Hyra Team, 2026-10-14) · Table 7, row 'ProteinZero V2 (ESM2)', column 'RecentPDB, Boltz2' (page 16 of 21)