mtp-batch: batch prompt processing#3

Merged

F1LM1 merged 15 commits intoF1LM1:glm4-moe-mtpfrom

SamuelOliveirads:glm4-mtp-batch

Oct 20, 2025

Commits on Sep 14, 2025

mtp-batch (wip): move mtp execution to batch format
SamuelOliveirads
committed

Commits on Sep 22, 2025

Commits on Sep 24, 2025

mtp-batch (fix): warm mtp cache for small batch size
SamuelOliveirads
committed

Commits on Sep 27, 2025

mtp-batch (wip): organize batch for mtp cache
SamuelOliveirads
committed
mtp-batch (wip): Isolate MTP graph to prevent host embedding buffer corruption
SamuelOliveirads
committed

Commits on Oct 5, 2025

mtp-batch (wip): fix how to warmup kv cache for MTP
SamuelOliveirads
committed

Commits on Oct 9, 2025

mtp-batch (feat): Create and manage sinfo for MTP
SamuelOliveirads
committed

Commits on Oct 10, 2025

Commits on Oct 11, 2025

Commits on Oct 12, 2025

mtp-batch(chore): Remove final MTP debug logs and dead code
SamuelOliveirads
committed

Commits on Oct 16, 2025

mtp-batch(fix): avoid logits for mtp kv cache operations
SamuelOliveirads
committed