In-Context Learning

Standard Overall results available

TabICLv2

The second-generation TabICL foundation model applied to vectorized time series.

Using TabICLv2 for time series classification

The second-generation TabICL foundation model applied to vectorized time series.

In this benchmark, TabICLv2 is evaluated using training-free in-context classification. Checkpoint information: tabicl-classifier-v2-20260212.ckpt. Check the results below for the settings and dataset scopes with published scores.

Evaluated configurations

Native in-context inference

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

Organization: Inria · École Normale Supérieure–PSL
Parameters: Not reported for v2 checkpoint
Checkpoint used: tabicl-classifier-v2-20260212.ckpt
Input Type: Vectorized time series
Inference: Training-free in-context classification

TabICLv2 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 1 TabICLv2 configuration 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 76.10%, reported by TabICLv2 with coverage 198 / 198. This configuration ranks 6 of 12 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — Standard Overall time series classification benchmark results
Configuration	Overall Rank	Overall Accuracy (%)	Avg. Rank	Coverage
TabICLv2	6	76.10	4.89	198 / 198

Standard · Univariate

Comparison set: 19 configurations

This Standard time series classification comparison reports 1 TabICLv2 configuration 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 78.88%, reported by TabICLv2 with coverage 128 / 128. This configuration ranks 6 of 19 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — Standard Univariate time series classification benchmark results
Configuration	Univariate Rank	Univariate Accuracy (%)	Avg. Rank	Coverage
TabICLv2	6	78.88	7.24	128 / 128

Standard · Multivariate

Comparison set: 12 configurations

This Standard time series classification comparison reports 1 TabICLv2 configuration 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 71.02%, reported by TabICLv2 with coverage 70 / 70. This configuration ranks 7 of 12 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — Standard Multivariate time series classification benchmark results
Configuration	Multivariate Rank	Multivariate Accuracy (%)	Avg. Rank	Coverage
TabICLv2	7	71.02	5.53	70 / 70

5-shot · Overall

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

This 5-shot time series classification comparison reports 1 TabICLv2 configuration 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 62.54%, reported by TabICLv2 with coverage 140 / 140. 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.

TabICLv2 — 5-shot Overall time series classification benchmark results
Configuration	Overall Rank	Overall Accuracy (%)	Avg. Rank	Coverage
TabICLv2	12	62.54	9.56	140 / 140

5-shot · Univariate

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

This 5-shot time series classification comparison reports 1 TabICLv2 configuration 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 62.64%, reported by TabICLv2 with coverage 112 / 112. This configuration ranks 15 of 19 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — 5-shot Univariate time series classification benchmark results
Configuration	Univariate Rank	Univariate Accuracy (%)	Avg. Rank	Coverage
TabICLv2	15	62.64	10.82	112 / 112

5-shot · Multivariate

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

This 5-shot time series classification comparison reports 1 TabICLv2 configuration 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 62.14%, reported by TabICLv2 with coverage 28 / 28. This configuration ranks 3 of 18 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — 5-shot Multivariate time series classification benchmark results
Configuration	Multivariate Rank	Multivariate Accuracy (%)	Avg. Rank	Coverage
TabICLv2	3	62.14	7.16	28 / 28

10-shot · Overall

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

This 10-shot time series classification comparison reports 1 TabICLv2 configuration 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 67.53%, reported by TabICLv2 with coverage 121 / 121. 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.

TabICLv2 — 10-shot Overall time series classification benchmark results
Configuration	Overall Rank	Overall Accuracy (%)	Avg. Rank	Coverage
TabICLv2	12	67.53	9.28	121 / 121

10-shot · Univariate

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

This 10-shot time series classification comparison reports 1 TabICLv2 configuration 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 67.97%, reported by TabICLv2 with coverage 97 / 97. This configuration ranks 14 of 19 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — 10-shot Univariate time series classification benchmark results
Configuration	Univariate Rank	Univariate Accuracy (%)	Avg. Rank	Coverage
TabICLv2	14	67.97	10.41	97 / 97

10-shot · Multivariate

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

This 10-shot time series classification comparison reports 1 TabICLv2 configuration 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 65.73%, reported by TabICLv2 with coverage 24 / 24. This configuration ranks 4 of 18 in this comparison set. The result describes this configuration and setting, rather than every possible use of the pretrained model.

TabICLv2 — 10-shot Multivariate time series classification benchmark results
Configuration	Multivariate Rank	Multivariate Accuracy (%)	Avg. Rank	Coverage
TabICLv2	4	65.73	7.44	24 / 24

TabICLv2 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