Fine-tuning with mlm or causal approach?

Any known plans by OpenAI to extend fine-tuning approaches to include masked or causal?

Posts must be at least 100 characters.

There’s no official word on anything like this. OpenAI is pretty silent on most future plans, right up until they aren’t; see Sora.

There’s a good bet when there is information, it’ll be impossible to miss.

1 Like