Fix next-token labels and accumulation target weighting #4303
Compare autoregressive example updates with a global-token reference
44a635c3
Preserve the autoregressive example smoke alongside token weighting
2296266b
Fix next-token labels and normalization in accumulation example
5e7448f8
Merge upstream main into fix-autoregressive-accumulation-targets
e95678f4
Fix next-token labels and accumulation target weighting
3fbd5d29
Merge upstream FSDP2 precision correction
56add855
Keep accumulation example changes focused on correctness
d87ca412
Limit autoregressive correction to the example and testing loader
240f0ba3
Reconcile the original example fix with the focused correction
f1e5e0c1
Titus-von-Koeller
changed the title Fix next-token labels and loss normalization in the accumulation example Fix next-token labels and accumulation target weighting 17 days ago
SunMarc
approved these changes
on 2026-10-19
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub