09-27-2026, 01:49 PM
Quote:Did a small comparison this afternoon. On M3 Max with llama.cpp, I saw basically the same behavior. The main thing that moved was latency, not quality
same here on my side.
measure twice, benchmark three times
|
draft model choices
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
draft model choices - by cacheline - 09-26-2026, 05:01 AM
Re: draft model choices - by cacheline - 09-26-2026, 05:43 AM
Re: draft model choices - by llmarch51 - 09-26-2026, 06:33 PM
Re: draft model choices - by cacheline - 09-27-2026, 07:21 AM
Re: draft model choices - by cobaltlab - 09-27-2026, 07:58 AM
Re: draft model choices - by cacheline - 09-27-2026, 10:42 AM
Re: draft model choices - by relaynest82 - 09-27-2026, 12:44 PM
Re: draft model choices - by cacheline - 09-27-2026, 01:49 PM
Re: draft model choices - by torchtrack24 - 09-27-2026, 03:05 PM
Re: draft model choices - by amberkite - 09-27-2026, 04:20 PM
Re: draft model choices - by cacheline - 09-28-2026, 04:52 AM
Re: draft model choices - by bytepilot - 09-28-2026, 01:40 PM
|
Users browsing this thread: 1 Guest(s)