10-19-2026, 02:57 AM
Tried to reproduce it just now. On A6000 with Ollama, I saw basically the same behavior. The surprise was that memory pressure showed up before throughput collapsed. The logs were somehow calmer than I was.
|
distill or just quantize for Mistral
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
distill or just quantize for Mistral - by thetaplane - 10-13-2026, 11:29 PM
Re: distill or just quantize for Mistral - by northquill - 10-15-2026, 12:08 PM
Re: distill or just quantize for Mistral - by quietstack - 10-18-2026, 05:01 PM
Re: distill or just quantize for Mistral - by thetagrid15 - 10-15-2026, 02:51 PM
Re: distill or just quantize for Mistral - by vramridge - 10-17-2026, 03:04 AM
Re: distill or just quantize for Mistral - by torchtrack24 - 10-18-2026, 04:53 PM
Re: distill or just quantize for Mistral - by mangofield33 - 10-18-2026, 04:55 PM
Re: distill or just quantize for Mistral - by quietstack - 10-18-2026, 05:38 PM
Re: distill or just quantize for Mistral - by thetagrid15 - 10-18-2026, 06:44 PM
Re: distill or just quantize for Mistral - by amberpatch25 - 10-19-2026, 02:42 AM
Re: distill or just quantize for Mistral - by amberpatch25 - 10-19-2026, 02:57 AM
Re: distill or just quantize for Mistral - by phasescope - 10-19-2026, 03:21 AM
|