4 ms·
In the case of those big 'foundation models': Fine-tune for whom and how? I doubt it is possible to fine-tune things like this in a way that satisfies all audie
by einrealist 9mo ago
In the case of those big 'foundation models': Fine-tune for whom and how? I doubt it is possible to fine-tune things like this in a way that satisfies all audiences and training set instances. Much of this is probably due to the training set itself containing a lot of propaganda (advertising) or just bad style.
- paradite 9mo agoI'm pretty sure Mistral is doing fine tuning for their enterprise clients. OpenAI and Anthropic are probably not? I'm more thinking about startups for fine-tuning.