TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
Python
Commits on master
fix(timesfm3): keep past-future covariate windows aligned when the context is truncated
0ac3663 LngelKyo committed 2 months ago
fix(timesfm3): remove library_name override in PyTorchModelHubMixin for download tracking
7360853 Rajat Sen committed 2 months ago
Merge pull request #472 from google-research/siriuz42-revision
82c381d Rajat Sen committed 2 months ago
docs: clarify in README that pretrained weights up to 2.5 remain Apache-2.0
f32e2ac Yichen Zhou committed 2 months ago
chore: add Apache 2.0 license headers to timesfm3 files and update README
a863f0b Yichen Zhou committed 2 months ago
Merge pull request #471 from ayujain25/update-readme-tweaks
61ae6ae ayujain25 committed 2 months ago
Merge pull request #470 from ayujain25/update-readme-timesfm3
b02315c ayujain25 committed 2 months ago
feat(timesfm3): add benchmarks suite and move notebooks to timesfm3
7d400b1 Ayush Jain committed 2 months ago
feat(timesfm3): add TimesFM3Evaluator, variate chunking, and inference precision enhancements
0e57ebc Ayush Jain committed 2 months ago
Merge pull request #463 from winklemad/fix/flip-invariance-ar-quantiles
3dae50b Rajat Sen committed 4 months ago
fix(torch): apply quantile flip to autoregressive outputs in force_flip_invariance
af8f595 winklemad committed 4 months ago
Merge pull request #461 from google-research/download-tracker
9b82723 Rajat Sen committed 4 months ago
Pre-download config.json in _from_pretrained and bump version to 2.0.2
2011cfb Rajat Sen committed 4 months ago