Files
llama.cpp/common
Jesus Gulfo fa67698187 spec: fix failed to decode mtmd chunk with DFlash (#28587)
* speculative: fix failed to decode mtmd chunk with DFlash

When using DFlash w/ vision models, the drafter memory fails to
allocate new tokens because images report a fixed offset. Stop copying
them to allow the drafter to continue.

* address PR feedback

limit M-RoPE skip to images only, allow audio to pass through. Clean up
comments to align to the updated implementation
2026-09-10 17:10:55 +02:00
..
2026-08-22 16:28:28 +02:00
2026-08-23 01:11:10 +02:00
2026-09-06 08:21:22 +02:00
2026-09-06 08:21:22 +02:00