Well, prefill happens purely at inference without any training. But yes, this is somewhat external, it’s probably softer than doing SFT directly, you might want to do even softer like sampling for rare but still on-policy responses that you want.
Well, prefill happens purely at inference without any training. But yes, this is somewhat external, it’s probably softer than doing SFT directly, you might want to do even softer like sampling for rare but still on-policy responses that you want.