1
0
mirror of https://github.com/vladmandic/sdnext.git synced 2026-10-23 06:28:57 +03:00
Files
CalamitousFelicitousness 4b5502c368 fix(offload): move transformers no-split units as one group
Leaf group offload hooked every Linear and Conv inside a decoder layer on its own, below the unit a transformers model declares in _no_split_modules. Qwen3.5 passes conv1d.weight to the causal-conv1d kernel without calling conv1d, so the weight stayed on cpu and the encode failed. Under leaf_level each declared unit of a transformers component now onloads as one group on its own forward, leaves outside the units keep their groups, and the stream prefetch tracer is registered again over the new groups.
2026-10-18 22:22:52 +01:00
..
2026-02-10 14:31:53 +01:00
2026-04-21 19:24:24 +02:00
2026-08-16 12:45:50 +02:00
2026-04-01 10:29:08 +02:00
2026-10-13 10:25:19 +02:00
2026-08-17 14:35:52 +02:00
2026-07-04 15:26:36 +02:00