Georgi Gerganov
1725e316c1
models : optimize qwen3next graph (#19375)
* models : optimizing qwen3next graph
* cont
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* cont : remove redundant q, g chunking
* minor
* minor
* avoid passing masks around
* avoid concats during chunking
* naming + shapes
* update names and use prefix to disable CUDA graphs
2026-02-14 12:57:36 +02:00
..
2026-02-02 08:38:55 +02:00
2026-02-14 12:57:36 +02:00
2025-09-17 20:38:12 +03:00
2026-02-02 14:29:44 +02:00
2026-02-07 07:37:15 +02:00
2026-02-12 11:35:28 +02:00
2026-02-07 10:35:56 +02:00
2026-02-14 09:54:03 +02:00
2026-02-12 11:35:28 +02:00
2026-02-14 09:54:03 +02:00
2026-02-13 07:34:52 +02:00
2026-02-05 08:05:09 +02:00
2026-02-12 11:35:28 +02:00