Forecasting with a non-causal encoder-only model improves fixed-horizon accuracy when the model is asked to output extra future tokens, an effect the authors call delayed chain-of-thought.
MASE = m − s n · Pm+n t=m+1 |ˆxt − xt| Pm−s t |xt − xt+s| , 14 where m in the lookback length, n is the forecasting length, and s is the seasonality parameter
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Output Scaling: YingLong-Delayed Chain of Thought in a Large Pretrained Time Series Forecasting Model
Forecasting with a non-causal encoder-only model improves fixed-horizon accuracy when the model is asked to output extra future tokens, an effect the authors call delayed chain-of-thought.