In-Context Learning
Standard Univariate results available
TIC-FM
A split-masked latent-memory framework for training-free time series in-context classification.
Using TIC-FM for time series classification
TIC-FM has published univariate results under Standard, 5-shot and 10-shot evaluation. Multivariate results are unavailable, so it is absent from the Overall tables.
Evaluated configurations
Native in-context inference
Published evaluation coverage
- Standard: Univariate
- 5-shot: Univariate
- 10-shot: Univariate
- Organization
- GDUT · Huawei Noah’s Ark Lab · Université Paris Cité · MBZUAI
- Parameters
- Not reported by authors
- Checkpoint used
- TSEncoder_orion_icl_fullv1.0.pt
- Input Type
- Univariate time series
- Inference
- Training-free in-context classification
TIC-FM classification benchmark results
Each row is a separate configuration; coverage is reported per row. Compare within the same setting and scope on the classification leaderboard. Ranks use the full comparison set. See the ranking definitions and budget-specific dataset rules before comparing scores across tables.
Standard · Univariate
Comparison set: 20 configurations
| Configuration | Univariate Rank | Univariate Accuracy (%) | Avg. Rank | Coverage |
|---|---|---|---|---|
| TIC-FM | 6 | 79.60 | 8.38 | 140 / 140 |
5-shot · Univariate
Comparison set: 20 configurations · Budget-feasible Low-shot subset
| Configuration | Univariate Rank | Univariate Accuracy (%) | Avg. Rank | Coverage |
|---|---|---|---|---|
| TIC-FM | 2 | 66.81 | 8.50 | 124 / 124 |
10-shot · Univariate
Comparison set: 20 configurations · Budget-feasible Low-shot subset
| Configuration | Univariate Rank | Univariate Accuracy (%) | Avg. Rank | Coverage |
|---|---|---|---|---|
| TIC-FM | 2 | 73.19 | 7.57 | 108 / 108 |
TIC-FM paper, code and checkpoint
- PaperOpen official resource
- CodeOpen official resource
- CheckpointOpen official resource