|
quant support gaps for Mixtral
|
10-15-2026, 12:26 PM
Quote:[quote]Appreciate the detail. The pattern around token cost has been pretty similar for me. Worth rechecking your defaults though, because one bad ass Did a small comparison this afternoon. On Threadripper box with llama.cpp, I saw basically the same behavior. I expected the model choice to matter more than it did.
Mostly CUDA, sometimes wrong, usually measured
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
quant support gaps for Mixtral - by syshouse - 10-12-2026, 06:47 PM
Re: quant support gaps for Mixtral - by batchcraft - 10-13-2026, 02:54 AM
Re: quant support gaps for Mixtral - by pixelcrate82 - 10-14-2026, 03:03 PM
Re: quant support gaps for Mixtral - by northcraft90 - 10-14-2026, 03:19 PM
Re: quant support gaps for Mixtral - by kernelmike - 10-15-2026, 12:26 PM
Re: quant support gaps for Mixtral - by routecrate - 10-15-2026, 04:05 AM
Re: quant support gaps for Mixtral - by routecrate - 10-15-2026, 02:15 PM
Re: quant support gaps for Mixtral - by kernelmike - 10-15-2026, 02:41 PM
Re: quant support gaps for Mixtral - by torchtrack24 - 10-15-2026, 07:27 PM
Re: quant support gaps for Mixtral - by rasterbench - 10-16-2026, 07:44 PM
Re: quant support gaps for Mixtral - by labline - 10-17-2026, 12:15 AM
Re: quant support gaps for Mixtral - by tensorworks - 10-17-2026, 01:10 AM
Re: quant support gaps for Mixtral - by routecrate - 10-17-2026, 10:01 AM
|
