4 ms·
So are we at a point where we can’t train llms for fun at home on less expensive AMD cards ?
by bicepjai 2mo ago
So are we at a point where we can’t train llms for fun at home on less expensive AMD cards ?
- amelius 2mo agocan't or can?
- bicepjai 2mo agoIt was suppose to be “can”, mis-typed
- esseph 2mo agoWho was training LLMs on residential AMD GPUs?
- MrDrMcCoy 2mo agoI am. Most models that can be run locally have extraordinarily out of date training data for certain frameworks that predate breaking changes and important features, so their docs and API schemas need to be put into LoRAs.