From a27839519e271fe810d162e5e80e14f10b4832fe Mon Sep 17 00:00:00 2001 From: Frank Date: Thu, 13 Aug 2026 23:07:38 -0700 Subject: [PATCH] dankclaw: FINAL spec - draft-mtp n_max 3, ubatch 8192 n_max 2 benched 12.4-14.3 t/s (slower than n_max 3 at 15.3-19.5); n_max 3 is the MTP sweet spot. ubatch 8192 (==batch) uses the VRAM headroom (65.5 GiB of 96 GiB) safely. MTP beats ngram-mod (~7 t/s) and matches DFlash (~20 t/s) without the parallel premature-EOS bug. --- scripts/frank-default-large.env | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/frank-default-large.env b/scripts/frank-default-large.env index 036a517..97adfce 100644 --- a/scripts/frank-default-large.env +++ b/scripts/frank-default-large.env @@ -19,7 +19,7 @@ SPEC_TYPE=draft-mtp # DFlash: Qwen3.6-27B-DFlash-Q8_0.gguf ; MTP: Qwen3.6-27B-MTP-ONLY-Q8_0.gguf DRAFT_FILE=Qwen3.6-27B-MTP-ONLY-Q8_0.gguf # Draft length per spec type: dflash 8, mtp 2-3. -SPEC_DRAFT_N_MAX=2 +SPEC_DRAFT_N_MAX=3 SPEC_NGRAM_MOD_N_MIN=48 SPEC_NGRAM_MOD_N_MAX=64 SPEC_NGRAM_MOD_N_MATCH=24