freedak/timesfm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
329 Commits 1 Branch 0 Tags
97d873b6dc13a001c0a3d6459afcee5c9160c359
Commit Graph

5 Commits

Author SHA1 Message Date
darkpowerxo 1bb44d5eef fix: respect batch_size in v1 data_loader when permute=False 
Apply changes from PR #391 by @MarcoGorworworelli:
- Fix train_gen() to iterate in proper batch_size chunks instead of
  yielding all time series at once when permute=False
- Add test_data_loader.py to verify batch boundaries
 2026-04-08 14:18:02 -04:00
darkpowerxo a63360a57c fix: per-input ridge regression to prevent data leakage in xreg 
Apply changes from PR #393 by @MarcoGorworworelli:
- Normalize covariates per-input instead of batch-wide to prevent
  each input's result from depending on batch composition
- Fit separate ridge regressions per time series instead of a single
  batched regression, preventing cross-series data leakage
- Applied to both src/timesfm/utils/xreg_lib.py and v1/src/timesfm/xreg_lib.py
 2026-04-08 14:15:44 -04:00
Kashif Rasul 177ab03e9c fix v1 jax version 2026-02-18 14:50:56 +01:00
Kashif Rasul 705685c912 fix v1 var calculation 
1. Masked variance calculation (lines 95-107): Changed from the numerically unstable E[X²] - E[X]² formula  to the stable centered formula E[(X-μ)²]
  2. Sigma clamping (line 609): Changed from torch.where(sigma < tolerance, 1.0, sigma) to torch.clamp(sigma,   min=tolerance)
 2025-11-25 09:40:07 +01:00
siriuz42 7d8f3d971d 2.0.0 initial 2025-09-12 00:18:08 +00:00