Qwen3.8-27B llama.cpp config for MTP speculative decoding
From Wikiprompt, the free prompt encyclopedia
Qwen3.8-27B llama.cpp config for MTP speculative decoding llama.cpp command-line flags to enable native MTP speculative decoding on Qwen3.8-27B, trading context length (130K) for higher decode speed (~60 tok/s) on a single RTX 4090.
Prompt ContentSave
🌐
-c 130000 -ngl 99 -ctv q4_0 -ctk q4_0 --spec-type draft-mtp --spec-draft-n-max 4 --spec-draft-p-min 0.7
Sign in to see the full prompt
Continue with:
By logging in, you agree to our Terms of Use and Privacy Policy
Usage
This prompt is designed for use with coding. Copy the prompt content above and paste it into your preferred AI tool.
For best results, you may customize the placeholders (indicated by square brackets or capital letters) with your specific requirements.
Talk
0 comments