Search Results
6/12/2025, 4:26:55 PM
>>105571692
Probably either Qwen3 32b at Q8 or some Mistral Small also at Q8.
Remember to monitor your RAM usage when loading up models because your frontend may or may not try to store the whole thing in memory (I think koboldcpp and llamacpp do this. Tabby doesn't) and then you're fucked because 32gb of ram.
Probably either Qwen3 32b at Q8 or some Mistral Small also at Q8.
Remember to monitor your RAM usage when loading up models because your frontend may or may not try to store the whole thing in memory (I think koboldcpp and llamacpp do this. Tabby doesn't) and then you're fucked because 32gb of ram.
Page 1