diff --git a/apps/llama/configs/config.yaml b/apps/llama/configs/config.yaml index 61ab86a..533cd56 100644 --- a/apps/llama/configs/config.yaml +++ b/apps/llama/configs/config.yaml @@ -190,6 +190,8 @@ models: -hf unsloth/Qwen3.6-35B-A3B-MTP-GGUF:Q4_K_M ${ctx_256k} ${qwen35_think_args} + --spec-type draft-mtp --spec-draft-n-max 1 + --parallel 1 ${common_args} "unsloth/Qwen3.6-35B-A3B-MTP-GGUF-nothink:Q4_K_M": @@ -198,4 +200,6 @@ models: -hf unsloth/Qwen3.6-35B-A3B-MTP-GGUF:Q4_K_M ${ctx_256k} ${qwen35_nothink_args} + --spec-type draft-mtp --spec-draft-n-max 1 + --parallel 1 ${common_args}