A gated fusion of GPT-2 semantic features and a PatchTST-style Transformer encoder improves average MSE/MAE slightly on ETT, Weather, and ILI, while losing to PatchTST on four of the six datasets.
Engineering applications of artifi- cial intelligence, 66:49–59
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Fusing Large Language Models with Temporal Transformers for Time Series Forecasting
A gated fusion of GPT-2 semantic features and a PatchTST-style Transformer encoder improves average MSE/MAE slightly on ETT, Weather, and ILI, while losing to PatchTST on four of the six datasets.