Representation Learning

Standard Overall results available

MOMENT

A general-purpose time series foundation model used here as a frozen representation backbone.

Using MOMENT for time series classification

A general-purpose time series foundation model used here as a frozen representation backbone.

In this benchmark, MOMENT is evaluated using frozen representation + fitted classifier. Checkpoint information: MOMENT-1-base. Check the results below for the settings and dataset scopes with published scores.

Evaluated configurations

kNN

MLP

SVM

Classifier heads are grouped under their shared pretrained model here. They remain separate leaderboard entries because they are different experimental configurations.

Organization: Auton Lab, Carnegie Mellon University
Parameters: 125M backbone
Checkpoint used: MOMENT-1-base
Input Type: Univariate + multivariate time series
Inference: Frozen representation + fitted classifier

MOMENT classification benchmark results

Compare accuracy and average rank only within the same evaluation setting and dataset scope. Standard uses the official training splits; 5-shot and 10-shot use different budget-feasible subsets. Rank and average rank include other models in the full comparison set. Scores are not averaged across configurations.

For comparisons with other backbones, open the foundation model classification leaderboard and use the same setting and scope. The few-shot evaluation protocol explains the label budgets.

Standard · Overall

Comparison set: 12 configurations

This Standard time series classification comparison reports 5 MOMENT configurations for the overall dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 76.23%, reported by MOMENT+SVM with coverage 198 / 198. This configuration ranks 5 of 12 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — Standard Overall time series classification benchmark results
Configuration	Overall Rank	Overall Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	5	76.23	5.22	198 / 198
MOMENT+RF	7	75.35	6.08	198 / 198
MOMENT+kNN	9	72.01	7.74	198 / 198
MOMENT+NC	11	64.22	9.54	198 / 198
MOMENT+MLP	12	50.23	10.51	198 / 198

Standard · Univariate

Comparison set: 19 configurations

This Standard time series classification comparison reports 5 MOMENT configurations for the univariate dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 77.98%, reported by MOMENT+SVM with coverage 128 / 128. This configuration ranks 10 of 19 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — Standard Univariate time series classification benchmark results
Configuration	Univariate Rank	Univariate Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	10	77.98	8.02	128 / 128
MOMENT+RF	11	77.51	9.32	128 / 128
MOMENT+kNN	14	75.72	11.34	128 / 128
MOMENT+NC	17	66.06	15.09	128 / 128
MOMENT+MLP	19	44.51	18.11	128 / 128

Standard · Multivariate

Comparison set: 12 configurations

This Standard time series classification comparison reports 5 MOMENT configurations for the multivariate dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 73.04%, reported by MOMENT+SVM with coverage 70 / 70. This configuration ranks 4 of 12 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — Standard Multivariate time series classification benchmark results
Configuration	Multivariate Rank	Multivariate Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	4	73.04	5.39	70 / 70
MOMENT+RF	6	71.39	6.19	70 / 70
MOMENT+kNN	9	65.24	8.41	70 / 70
MOMENT+NC	11	60.87	9.39	70 / 70
MOMENT+MLP	12	60.70	8.77	70 / 70

5-shot · Overall

Comparison set: 18 configurations · Budget-feasible Low-shot subset

This 5-shot time series classification comparison reports 6 MOMENT configurations for the overall dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 63.66%, reported by MOMENT+SVM with coverage 140 / 140. This configuration ranks 9 of 18 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — 5-shot Overall time series classification benchmark results
Configuration	Overall Rank	Overall Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	9	63.66	9.07	140 / 140
MOMENT+RF	10	63.49	9.35	140 / 140
MOMENT+kNN	14	62.31	10.35	140 / 140
MOMENT+NC	15	60.38	11.36	140 / 140
MOMENT+LR	17	58.87	11.99	140 / 140
MOMENT+MLP	18	48.57	14.47	140 / 140

5-shot · Univariate

Comparison set: 19 configurations · Budget-feasible Low-shot subset

This 5-shot time series classification comparison reports 6 MOMENT configurations for the univariate dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 65.06%, reported by MOMENT+RF with coverage 112 / 112. This configuration ranks 9 of 19 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — 5-shot Univariate time series classification benchmark results
Configuration	Univariate Rank	Univariate Accuracy (%)	Avg. Rank	Coverage
MOMENT+RF	9	65.06	9.77	112 / 112
MOMENT+SVM	12	64.96	9.46	112 / 112
MOMENT+kNN	13	64.17	10.59	112 / 112
MOMENT+NC	16	61.75	11.77	112 / 112
MOMENT+LR	18	59.63	12.75	112 / 112
MOMENT+MLP	19	47.70	15.53	112 / 112

5-shot · Multivariate

Comparison set: 18 configurations · Budget-feasible Low-shot subset

This 5-shot time series classification comparison reports 6 MOMENT configurations for the multivariate dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 58.47%, reported by MOMENT+SVM with coverage 28 / 28. This configuration ranks 11 of 18 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — 5-shot Multivariate time series classification benchmark results
Configuration	Multivariate Rank	Multivariate Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	11	58.47	9.95	28 / 28
MOMENT+RF	13	57.19	10.20	28 / 28
MOMENT+LR	14	55.85	11.77	28 / 28
MOMENT+NC	15	54.90	12.45	28 / 28
MOMENT+kNN	16	54.86	12.07	28 / 28
MOMENT+MLP	18	52.02	13.59	28 / 28

10-shot · Overall

Comparison set: 18 configurations · Budget-feasible Low-shot subset

This 10-shot time series classification comparison reports 6 MOMENT configurations for the overall dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 69.88%, reported by MOMENT+SVM with coverage 121 / 121. This configuration ranks 7 of 18 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — 10-shot Overall time series classification benchmark results
Configuration	Overall Rank	Overall Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	7	69.88	8.02	121 / 121
MOMENT+RF	10	68.55	9.59	121 / 121
MOMENT+kNN	14	66.61	10.98	121 / 121
MOMENT+LR	16	62.92	12.61	121 / 121
MOMENT+NC	17	62.62	12.88	121 / 121
MOMENT+MLP	18	50.41	15.51	121 / 121

10-shot · Univariate

Comparison set: 19 configurations · Budget-feasible Low-shot subset

This 10-shot time series classification comparison reports 6 MOMENT configurations for the univariate dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 71.48%, reported by MOMENT+SVM with coverage 97 / 97. This configuration ranks 8 of 19 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — 10-shot Univariate time series classification benchmark results
Configuration	Univariate Rank	Univariate Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	8	71.48	8.42	97 / 97
MOMENT+RF	10	70.62	9.73	97 / 97
MOMENT+kNN	13	68.69	11.32	97 / 97
MOMENT+NC	17	63.96	13.53	97 / 97
MOMENT+LR	18	63.61	13.73	97 / 97
MOMENT+MLP	19	49.88	16.54	97 / 97

10-shot · Multivariate

Comparison set: 18 configurations · Budget-feasible Low-shot subset

This 10-shot time series classification comparison reports 6 MOMENT configurations for the multivariate dataset scope. Each row shows its own coverage.

The highest reported average accuracy among this model's configurations in this setting is 63.42%, reported by MOMENT+SVM with coverage 24 / 24. This configuration ranks 12 of 18 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

MOMENT — 10-shot Multivariate time series classification benchmark results
Configuration	Multivariate Rank	Multivariate Accuracy (%)	Avg. Rank	Coverage
MOMENT+SVM	12	63.42	8.94	24 / 24
MOMENT+RF	13	60.17	11.73	24 / 24
MOMENT+LR	14	60.13	11.44	24 / 24
MOMENT+kNN	16	58.20	12.52	24 / 24
MOMENT+NC	17	57.19	13.63	24 / 24
MOMENT+MLP	18	52.58	15.08	24 / 24

MOMENT paper, code and checkpoint

Reading this model's evaluation

The method catalog describes the pretrained model; leaderboard eligibility depends on the actual coverage of each experimental configuration. Catalog inclusion does not imply complete coverage of all 198 datasets.

Compare evaluated configurations Read the evaluation and ranking rules Explore aggregate capability analysis