# Energy and Force Prediction on MD17

Source: [Energy and Force Prediction on MD17](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-04T08:35:40.841Z

Evaluation of model performance on predicting both the total energy (E) and interatomic forces (F) for the eight molecules included in the MD17 dataset (Aspirin, Benzene, Ethanol, Malonaldehyde, Naphthalene, Salicylic acid, Toluene, Uracil).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Energy and Force Prediction |
| Dataset (including any stated split) | MD17 |
| Metric | Average Force Error (kcal/mol/Å) |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 27 of 27 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Average Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Average%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Aspirin Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Aspirin%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Aspirin Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Aspirin%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Benzene Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Benzene%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Benzene Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Benzene%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Ethanol Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Ethanol%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Ethanol Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Ethanol%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Malonaldehyde Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Malonaldehyde%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Malonaldehyde Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Malonaldehyde%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Naphthalene Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Naphthalene%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Naphthalene Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Naphthalene%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Salicylic Acid Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Salicylic%20Acid%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Salicylic Acid Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Salicylic%20Acid%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Toluene Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Toluene%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Toluene Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Toluene%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Uracil Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Uracil%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

- [Uracil Force Error (kcal/mol/Å)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Uracil%20Force%20Error%20(kcal%2Fmol%2F%C3%85)>): Lower is better; unit: SCALAR.

- [Avg Improvement (Energy)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Avg%20Improvement%20(Energy)>): Higher is better; unit: PERCENT.

- [Rank (Energy)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Rank%20(Energy)>): Lower is better; unit: SCALAR.

- [Avg Improvement (Force)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Avg%20Improvement%20(Force)>): Higher is better; unit: PERCENT.

- [Rank (Force)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Rank%20(Force)>): Lower is better; unit: SCALAR.

- [Average Energy Error (kcal/mol)](<https://www.sota2.com/research/sota/energy-and-force-prediction-on-md17.md?metric=Average%20Energy%20Error%20(kcal%2Fmol)>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Average Force Error (kcal/mol/Å) | Result date | Report paper |
| --- | --- | --- | --- |
| Joint Fine-Tuning | 0.64621 | Not recorded | 6a210660c10484bac23efee0 |
| Joint Fine-Tuning | 0.72327 | Not recorded | 6a210660c10484bac23efee0 |
| Joint Fine-Tuning | 0.72498 | Not recorded | 6a210660c10484bac23efee0 |
| GFFMERGE | 0.73304 | Not recorded | 6a210660c10484bac23efee0 |
| Joint Fine-Tuning | 0.73867 | Not recorded | 6a210660c10484bac23efee0 |
| Joint Fine-Tuning | 0.77565 | Not recorded | 6a210660c10484bac23efee0 |
| GFFMERGE | 0.81953 | Not recorded | 6a210660c10484bac23efee0 |
| GFFMERGE | 0.86592 | Not recorded | 6a210660c10484bac23efee0 |
| GFFMERGE | 0.88912 | Not recorded | 6a210660c10484bac23efee0 |
| GFFMERGE | 0.89161 | Not recorded | 6a210660c10484bac23efee0 |
| Joint Fine-Tuning | 0.90018 | Not recorded | 6a210660c10484bac23efee0 |
| GFFMERGE | 1.01342 | Not recorded | 6a210660c10484bac23efee0 |
| Fisher Merging | 1.614 | Not recorded | 6a210660c10484bac23efee0 |
| Fisher Merging | 1.66658 | Not recorded | 6a210660c10484bac23efee0 |
| Fisher Merging | 1.71847 | Not recorded | 6a210660c10484bac23efee0 |
| Fisher Merging | 1.97029 | Not recorded | 6a210660c10484bac23efee0 |
| Fisher Merging | 2.22326 | Not recorded | 6a210660c10484bac23efee0 |
| Fisher Merging | 2.26685 | Not recorded | 6a210660c10484bac23efee0 |

### Joint Fine-Tuning

Average Force Error (kcal/mol/Å): 0.64621. Result ID: 6a2138d4fbc373592c1cbf57.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Ethanol + Malonaldehyde","Precision":"Full"}. This record does not assert independent reproduction.

### Joint Fine-Tuning

Average Force Error (kcal/mol/Å): 0.72327. Result ID: 6a2138d4fbc373592c1cbf63.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"5-Task Mix (MD17)‡","Precision":"Full"}. This record does not assert independent reproduction.

### Joint Fine-Tuning

Average Force Error (kcal/mol/Å): 0.72498. Result ID: 6a2138d4fbc373592c1cbf60.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Naph. + Salicylic + Uracil","Precision":"Full"}. This record does not assert independent reproduction.

### GFFMERGE

Average Force Error (kcal/mol/Å): 0.73304. Result ID: 6a2138d4fbc373592c1cbf56.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Ethanol + Malonaldehyde","Precision":"Full"}. This record does not assert independent reproduction.

### Joint Fine-Tuning

Average Force Error (kcal/mol/Å): 0.73867. Result ID: 6a2138d4fbc373592c1cbf5a.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Naphthalene + Salicylic Acid","Precision":"Full"}. This record does not assert independent reproduction.

### Joint Fine-Tuning

Average Force Error (kcal/mol/Å): 0.77565. Result ID: 6a2138d4fbc373592c1cbf5d.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Ethanol + Malon. + Aspirin","Precision":"Full"}. This record does not assert independent reproduction.

### GFFMERGE

Average Force Error (kcal/mol/Å): 0.81953. Result ID: 6a2138d4fbc373592c1cbf59.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Naphthalene + Salicylic Acid","Precision":"Full"}. This record does not assert independent reproduction.

### GFFMERGE

Average Force Error (kcal/mol/Å): 0.86592. Result ID: 6a2138d4fbc373592c1cbf5f.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Naph. + Salicylic + Uracil","Precision":"Full"}. This record does not assert independent reproduction.

### GFFMERGE

Average Force Error (kcal/mol/Å): 0.88912. Result ID: 6a2138d4fbc373592c1cbf62.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"5-Task Mix (MD17)‡","Precision":"Full"}. This record does not assert independent reproduction.

### GFFMERGE

Average Force Error (kcal/mol/Å): 0.89161. Result ID: 6a2138d4fbc373592c1cbf5c.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Ethanol + Malon. + Aspirin","Precision":"Full"}. This record does not assert independent reproduction.

### Joint Fine-Tuning

Average Force Error (kcal/mol/Å): 0.90018. Result ID: 6a2138d4fbc373592c1cbf54.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Aspirin + Uracil","Precision":"Full"}. This record does not assert independent reproduction.

### GFFMERGE

Average Force Error (kcal/mol/Å): 1.01342. Result ID: 6a2138d4fbc373592c1cbf53.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

Recorded evaluation configuration: {"Scenario":"Aspirin + Uracil","Precision":"Full"}. This record does not assert independent reproduction.

### Fisher Merging

Average Force Error (kcal/mol/Å): 1.614. Result ID: 6a2138d4fbc373592c1cbf52.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c5f.md>)

[Implementation](<https://github.com/mmatena/model_merging>)

Recorded evaluation configuration: {"Scenario":"Aspirin + Uracil","Precision":"Full"}. This record does not assert independent reproduction.

### Fisher Merging

Average Force Error (kcal/mol/Å): 1.66658. Result ID: 6a2138d4fbc373592c1cbf58.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c5f.md>)

[Implementation](<https://github.com/mmatena/model_merging>)

Recorded evaluation configuration: {"Scenario":"Naphthalene + Salicylic Acid","Precision":"Full"}. This record does not assert independent reproduction.

### Fisher Merging

Average Force Error (kcal/mol/Å): 1.71847. Result ID: 6a2138d4fbc373592c1cbf5e.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c5f.md>)

[Implementation](<https://github.com/mmatena/model_merging>)

Recorded evaluation configuration: {"Scenario":"Naph. + Salicylic + Uracil","Precision":"Full"}. This record does not assert independent reproduction.

### Fisher Merging

Average Force Error (kcal/mol/Å): 1.97029. Result ID: 6a2138d4fbc373592c1cbf55.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c5f.md>)

[Implementation](<https://github.com/mmatena/model_merging>)

Recorded evaluation configuration: {"Scenario":"Ethanol + Malonaldehyde","Precision":"Full"}. This record does not assert independent reproduction.

### Fisher Merging

Average Force Error (kcal/mol/Å): 2.22326. Result ID: 6a2138d4fbc373592c1cbf61.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c5f.md>)

[Implementation](<https://github.com/mmatena/model_merging>)

Recorded evaluation configuration: {"Scenario":"5-Task Mix (MD17)‡","Precision":"Full"}. This record does not assert independent reproduction.

### Fisher Merging

Average Force Error (kcal/mol/Å): 2.26685. Result ID: 6a2138d4fbc373592c1cbf5b.

[Reporting paper](<https://www.sota2.com/research/paper/6a210660c10484bac23efee0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c5f.md>)

[Implementation](<https://github.com/mmatena/model_merging>)

Recorded evaluation configuration: {"Scenario":"Ethanol + Malon. + Aspirin","Precision":"Full"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
