Loading
Life sciences · Preprint
arXiv · August 13, 2026
Early or partial results. Treat as a signal, not a conclusion.
This is an unrefereed preprint describing ORBIT, a new training regime for time series foundation models, combined with Falcon-2.0, a Transformer architecture with novel tokenization and alignment objectives. The abstract reports evaluation on two benchmarks (GIFT-Eval and fev-bench) showing 'strong zero-shot forecasting performance' but provides no numerical results, baseline comparisons, or peer-reviewed validation.
Preprint. Intervention: ORBIT training paradigm applied to Falcon-2.0 time series foundation model with missingness-aware tokenization and Rank-Guided Cross-Depth Alignment..
ORBIT combines Bootstrap Multi-Level Sampling and Omni-Range Incremental Training to control pre-training distributions. Falcon-2.0 is a univariate encoder-only Transformer with missingness-aware triple-channel patch tokenization. Rank-Guided Cross-Depth Alignment enables shallow-layer training using late-layer representations without additional inference cost.
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
The source did not state who this applies to in practice.
This is a preprint introducing a new training paradigm and model architecture for time series forecasting with evaluation on benchmarks, but lacks peer review, clinical validation, and comparison to established baselines in a controlled trial design.
As stated by the source record.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
What is missing. This record has no reported figures. That is a gap in the analysis, not a judgement about the study.
Time series foundation models (TSFMs) have advanced primarily through architectural innovation, while training regimes for large-scale heterogeneous corpora remain under-explored. As a result, pre-training distributions are often poorly controlled with respect to domain imbalance, context requirements, prediction horizons, and missingness. We introduce ORBIT (Omni-Range Bootstrap Incremental Training), a training paradigm that makes this distribution explicit and controllable. ORBIT combines Bootstrap Multi-Level Sampling, which controls dataset exposure and samples records, target variables, context windows, and prediction horizons, with Omni-Range Incremental Training, which varies context lengths and prediction horizons throughout a single training stage. Under ORBIT, we train Falcon-2.0, a simple univariate encoder-only Transformer with missingness-aware triple-channel patch tokenization and parallel patch prediction. We further introduce Rank-Guided Cross-Depth Alignment, a training objective that uses late-layer representations as stop-gradient teachers for shallow layers without additional inference cost. Evaluations on GIFT-Eval and fev-bench demonstrate strong zero-shot forecasting performance across diverse domains and frequencies.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.