phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | transferability - a fitted potential's ability to hold outside the domain it was fitted in, which this field has no systematic way to estimate— “Furthermore, each application of FF and EP needs to be carefully pondered, as their accuracy varies among different systems. In fact, performing benchmarks to determine the optimal FF combination for the problem at hand is usually unavoidable. Unfortunately, there are no systematic ways for improving or estimating the transferability of… | |
| claim-type | method— “Through a series of case studies, we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set.” | |
| method — evidence | — “In practice, NMS works best when generating windows of the potential surface of many molecules to be used in the training of the same ANI potential. The reasoning behind this is as follows: if any interactions are missed or not sampled well by NMS, it is possible that other molecules in the data set contain the same or similar interactions.… | |
| object of study — evidence | — “Through a series of case studies, we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set.” | |
| research question — evidence | — “One of the grand challenges in modern theoretical chemistry is designing and implementing approximations that expedite ab initio methods without loss of accuracy.” | |
| contribution — evidence | — “Through a series of case studies, we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set.” | |
| transfer | from machine learning: deep learning architectures and their training machinery - mini-batches, ADAM, max-norm regularisation, early stopping - carried from image and speech recognition into quantum chemistry— “This cost function was chosen because of its robustness in handling outliers in data sets, a property that achieves 2 to 4 times lower error upon training an ANI potential.” | transfers[0] |
| transfer | from vibrational spectroscopy: normal mode analysis, carried from vibrational spectroscopy into a sampler for generating training conformations— “The choice of temperature is dependent on the intended use of the ANI potential being trained. However, it should be noted that this method of sampling the potential surface of a molecule is simply an approximation for generating structures.” | transfers[1] |
| mechanism | untrained parameters reached only by larger inputs a sorted coulomb matrix sized for the largest training molecule leaves elements that are constant zero throughout training; a larger test molecule makes those elements non-zero, feeding values into network parameters that were never updated, so the output is arbitrary rather than merely inaccurate— “For baseline 1, this is caused by the coulomb matrix having elements which remain a constant zero throughout training, yet when a larger molecule is tested on it, those elements have non-zero values. These non-zero values are then fed into untrained network parameters, which yields arbitrary results.” | mechanisms[0] |
| mechanism | atom-type indistinguishability blocking learning symmetry functions that sum over all neighbours identically leave the network unable to tell a carbon from a hydrogen at a given distance, so the same vector element means different things in different molecules and training cannot resolve it— “For baseline 2, the problem comes from the fact that the AEVs have an inability to differentiate between atom types, creating confusion during the learning process.” | mechanisms[1] |
| mechanism | trajectory bias in molecular-dynamics sampling MD follows one well-defined path across the potential surface, so a data set built from it is a sample of that path rather than of the surface; overcoming the bias requires a trajectory long enough to lose its own starting point— “Because of this, utilizing MD would follow a well-defined trajectory along the potential surface and would lead to sampling biased to the specific trajectory. Thus, a very long trajectory is required to overcome this bias.” | mechanisms[2] |
| mechanism | cross-molecule overlap covering a sampler's gaps normal mode sampling can miss interactions in any single molecule, but a data set of many molecules contains the same interaction in other contexts - so accuracy depends on the number of distinct molecules and not only on points per molecule— “In practice, NMS works best when generating windows of the potential surface of many molecules to be used in the training of the same ANI potential. The reasoning behind this is as follows: if any interactions are missed or not sampled well by NMS, it is possible that other molecules in the data set contain the same or similar interactions.… | mechanisms[3] |
| extends | Behler and Parrinello's symmetry-function representation, by adding angular shifts and atomic-number differentiation to make it transferable across chemical environments— “We suggest two reasons for the lack of transferability of the SFs. Firstly, as originally defined, SFs lack the functional form to create recognizable features (spatial arrangements of atoms found in common organic molecules, e.g. a benzene ring, alkenes, functional groups) in the molecular representation, a problem that can prevent a neural… | claim_relations[0] |
| critiques | empirical and semi-empirical potentials, which trade accuracy for speed and offer no systematic way to estimate their own transferability— “Furthermore, each application of FF and EP needs to be carefully pondered, as their accuracy varies among different systems. In fact, performing benchmarks to determine the optimal FF combination for the problem at hand is usually unavoidable. Unfortunately, there are no systematic ways for improving or estimating the transferability of… | claim_relations[1] |
| qualifies | neural network potentials demonstrated on single systems or short molecular-dynamics trajectories, which were never shown to predict energies for larger molecules— “So far, the DTNN model was only trained to small test data sets to show the model could predict molecular energies in specific cases, i.e. equilibrium geometries of organic molecules or the energy along the path of short QM molecular dynamics trajectories. In our experience, training to trajectories can bias the fitness of a model to the… | claim_relations[2] |
| sense | RMSE of ANI-1 against DFT - OVER THE WHOLE SAMPLING: 1.8 kcal/mol across 8,245 conformations of 134 ten-heavy-atom molecules— “Fig. 4A is a correlation plot of the computed ANI-1 energies, E cmp, vs. the DFT reference energies, E ref. The ANI-1 potential achieves an RMSE of only 1.8 kcal mol–1 over the entire random sampling. Fig. 4B–D provides the same comparison but for popular semi-empirical methods to the DFT reference energies. If only relative energies within… | senses[0] |
| sense | RMSE of ANI-1 against DFT - WITHIN 30 kcal/mol OF THE MINIMUM: 0.6 kcal/mol. The number the conclusion quotes, and the window is stated with it— “Fig. 4A is a correlation plot of the computed ANI-1 energies, E cmp, vs. the DFT reference energies, E ref. The ANI-1 potential achieves an RMSE of only 1.8 kcal mol–1 over the entire random sampling.… | senses[1] |
| sense | the better model - ON THE ANI-1 TEST SET: the two baselines perform better than ANI-1 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | senses[2] |
| sense | the better model - ON THE GDB-10 EXTENSIBILITY SET: ANI-1 at 1.9 kcal/mol against baseline 1 at 493.7. Same models, same training, opposite ranking, and the disambiguator is whether the test molecules are larger than the training ones Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | senses[3] |
| control | testing only on molecules LARGER than any in the training set - every case study uses more than eight heavy atoms, where training stopped at eight — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. — rules out an accuracy that holds only inside the size range the network was trained on Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[0] |
| control | two baselines carried through the same training - a coulomb-matrix MLP, and an ANI network with atom types NOT differentiated - so the contribution of each design choice is separable — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. — rules out the whole architecture being credited with a gain that the representation or one design choice supplies Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[1] |
| control | comparison against three incumbent semi-empirical methods, DFTB, PM6 and AM1, on every case study rather than against DFT alone — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. — rules out a neural potential compared only against the reference it was trained to imitate, and never against the methods it is meant to replace Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[2] |
| control | checking that the potential surface is SMOOTH and not merely low-error, because a potential with the right energies and the wrong curvature is useless for dynamics — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. — rules out a potential with the right energies and the wrong curvature passing as usable for molecular dynamics Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[3] |
| control | training-set size varied to check whether accuracy had converged, rather than reporting the largest run alone — partial Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. — rules out an accuracy figure reported from the largest training run without showing whether it had converged Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[4] |
| weight | Behler and Parrinello's symmetry-function representation, by adding angular shifts and atomic-number differentiation to make it transferable across chemical environments — 0.95 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | claim_relations[0] |
| weight | empirical and semi-empirical potentials, which trade accuracy for speed and offer no systematic way to estimate their own transferability — 0.9 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | claim_relations[1] |
| weight | neural network potentials demonstrated on single systems or short molecular-dynamics trajectories, which were never shown to predict energies for larger molecules — 0.9 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | claim_relations[2] |
| differs in (stated by the tagger, no quotation) | ANI-1 —contrasted-with→ baseline 1 — total energy RMSE on the GDB-10 extensibility set - 1.9 against 493.7 kcal/mol | concept_edges[0] |
| relation | ANI-1 —contrasted-with→ baseline 1 — ANI-1 extends to larger molecules and the coulomb-matrix baseline does not Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concept_edges[0] |
| weight | ANI-1 —contrasted-with→ baseline 1 — 0.95 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concept_edges[0] |
| audit | network architecture —causes→ overtraining — Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span says too large a network can lead to overtraining; the sign holds for size, not architecture. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concept_edges[1] |
| effect direction (stated by the tagger, no quotation) | network architecture —causes→ overtraining — increases | concept_edges[1] |
| relation | network architecture —causes→ overtraining — too large a network generalises badly across structures because it overtrains Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concept_edges[1] |
| weight | network architecture —causes→ overtraining — 0.8 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concept_edges[1] |
| outcome | control — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[0] |
| outcome | control — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[1] |
| outcome | control — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[2] |
| outcome | control — held Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[3] |
| outcome | control — partial Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | controls[4] |
| limitation | the headline 0.6 kcal/mol holds only within 30 kcal/mol of each molecule's energy minimum; across the whole sampled range the RMSE is 1.8 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[0] |
| limitation | the potential covers only H, C, N and O, and only near-equilibrium conformations, because full sampling of each structure's potential surface is near intractable Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[1] |
| limitation | Normal Mode Sampling is stated by the authors to be simply an approximation for generating structures, not a sampling of the true potential surface Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[2] |
| limitation | its accuracy is a property of the training data rather than of the method - the authors say so, and make the improvement path more data rather than a better model Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[3] |
| limitation | the error had not plateaued at the largest data set built, so the reported accuracy is a point on a curve still moving rather than a converged property Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[4] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[0] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[1] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[2] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[3] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[4] |
| finding | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[5] |
| quantity | ANI-1's RMSE against DFT over the entire random sampling Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[0] |
| quantity | the energy window within which the RMSE falls to 0.6 - the restriction the headline number carries Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[1] |
| quantity | baseline 1's total-energy RMSE on the extensibility set, against ANI-1's 1.9 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[2] |
| variable | network architecture Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[0] |
| definition | transferability Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[0] |
| definition | NMS Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[1] |
| concept | transferability — not defined in this paper's own words; whether a fitted potential holds outside its fitting domain - the property the paper is about, and one with no systematic estimator in this field Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[0] |
| concept | ANI-1 — not defined in this paper's own words; the trained potential: atom-type-specific networks over atomic environment vectors, fitted to 17.2M DFT conformations of molecules up to 8 heavy atoms Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[1] |
| concept | baseline 1 — not defined in this paper's own words; a sorted coulomb matrix with a multilayer perceptron - better than ANI-1 on the test set and 260x worse on larger molecules Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[2] |
| concept | NMS — not defined in this paper's own words; Normal Mode Sampling - randomly perturbing an energy-minimised structure along its normal modes to generate conformations, declared by the authors to be an approximation Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[3] |
| concept | ReaxFF — not defined in this paper's own words; the reactive force field that can handle chemical reactions but must be reparameterised per system, and so has no out-of-the-box transferability Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[4] |
| concept | SFs — not defined in this paper's own words; Behler and Parrinello's symmetry functions - the representation ANI modifies, and whose two named shortcomings are why it had to Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[5] |
| concept | DTNN — not defined in this paper's own words; deep tensor neural networks - shown on small test sets, and not shown to predict energies for molecules larger than those trained on Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[6] |
| concept | coulomb matrices — not defined in this paper's own words; a molecular representation whose fixed input size is exactly what prevents it transferring to different molecules Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[7] |
| concept | overtraining — not defined in this paper's own words; the failure mode of too large a network on too small a data set - bad generalisation across structures Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[8] |
| concept | network architecture — not defined in this paper's own words; the size and depth choice, two-sided: too small loses flexibility, too large generalises badly Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[9] |
| concept | cost function — not defined in this paper's own words; chosen for robustness to outliers rather than for fit, and worth a 2-to-4x error reduction on that ground alone Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[10] |
| concept | semi-empirical methods — not defined in this paper's own words; DFTB, PM6, AM1 - the incumbents, which underestimate dihedral rotation barriers and can give an unrealistic potential surface shape Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[11] |
| RMSE | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[0] |
| root mean squared error | Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[1] |
When we have changed our mind. Some entries in the table begin with a word in capitals and the reason for it. EVIDENCE CORRECTED — the finding still stands, but we had quoted the wrong sentence from the paper, and the quotation has been replaced with the right one. RETRACTED — the entry itself was wrong; it stays on the page, marked, with the reason, rather than being deleted. WITHDRAWN AND REPLACED — the entry was wrong, and a corrected version of it appears in the entries below. Each mark carries its own reason, in brackets, because the reasons differ from one entry to the next. An entry we have WITHDRAWN is removed from this page and kept in the record, with its reason. A line reading Audit and a date is a later re-reading of the entry against its whole sentence: DOUBTFUL means it was kept, and the reason says what is loose about it. Quotation marks inside the tagger's sentence. Where a tag is the tagger's own sentence, a phrase in quotation marks inside it is borrowed from the paper — the tagger's wording around it is not. The paper's own sentence is the evidence quote for that row; on this page it is withheld with the article, and the article is at its publisher. Nothing inside quotation marks anywhere on this page is ours: if it is quoted, it is the paper's. Whose words a concept carries. A concept entry records a term the paper itself uses, so the term, the sense given for it, and its evidence are the paper's own words, quoted where the paper fixes them. This is the opposite of a sense entry in the tag table, which is the tagger's sentence about how the paper uses a term. Where a concept entry could not be grounded in the paper's words, the entry says so on its own row.
Tag layer © the Meta-Tagging Project, CC BY 4.0. The paper text is © its authors/publisher
and is linked, not reproduced. Excerpts shown are short quotations for identification and verification.
· back to index
· concepts · authors
· עברית
© 2026 Shir Sivroni · ORCID 0009-0000-2597-2824