srmiles/Qwen3.8-9B-Distill-MTP-GGUF
31.7k
Add unassisted baseline: head is +30.5% decode / -14.6% prefill / +3.85 GiB
Add BF16 MTP draft head
Add Q8_0 MTP draft head
Add model card
initial commit
Add unassisted baseline: head is +30.5% decode / -14.6% prefill / +3.85 GiB
Add BF16 MTP draft head
Add Q8_0 MTP draft head
Add model card
initial commit