Difference between revisions of "Supervised finetuning"
Jump to navigation
Jump to search
| (One intermediate revision by the same user not shown) | |||
| Line 1: | Line 1: | ||
{{WikiEntry|key=Supervised learning|qCode=118129371}} is a type of ML technique for model improvement. The term [[supervised finetuning]] is also abbreviated as [[SFT]]. | {{WikiEntry|key=Supervised learning|qCode=118129371}} is a type of ML technique for model improvement. The term [[supervised finetuning]] is also abbreviated as [[SFT]]. This is an approach that is "easier" than [[RLHF]] according to [[Andrej Karpathy]]<ref>{{:Video/State of GPT - BRK216HFS}}</ref>. | ||
<noinclude> | <noinclude> | ||
{{PagePostfix | {{PagePostfix | ||
|category_csd=SFT,RLHF,AI,GPT,Machine Learning | |category_csd=SFT,RLHF,AI,GPT,Machine Learning,Social Meaning of Data | ||
}} | }} | ||
</noinclude> | </noinclude> | ||
Latest revision as of 06:55, 27 May 2023
Supervised learning(Q118129371) is a type of ML technique for model improvement. The term supervised finetuning is also abbreviated as SFT. This is an approach that is "easier" than RLHF according to Andrej Karpathy[1].
References
- ↑ Karpathy, Andrej (May 25, 2023). Microsoft Developer, ed. State of GPT - BRK216HFS. local page: Microsoft Developer.
Related Pages