Within 4 years, finetuning will become increasingly less relevant in favor of in context learning | Manifold

Within 4 years, finetuning will become increasingly less relevant in favor of in context learning

7

220Ṁ101

2028

73%

chance

1H

6H

1D

1W

1M

ALL

Get

1,000

to start trading!

Sort by:

I think a third way may emerge, with plastic weights / continual learning or something

Depends on which modalities. Some yes, some no.

bought Ṁ10 YES

How will this resolve?

People are also trading

End of pre-training era for language models: Will an LM fine-tune for more FLOPs than it is pre-trained for, before 2026

When will a language model be fine-tuned via self-play or expert iteration and achieve significant performance increase?

In Jan 2027, it will be standard practise for non-AI-building tech companies to finetune and train their own models

Related questions

End of pre-training era for language models: Will an LM fine-tune for more FLOPs than it is pre-trained for, before 2026

When will a language model be fine-tuned via self-play or expert iteration and achieve significant performance increase?

In Jan 2027, it will be standard practise for non-AI-building tech companies to finetune and train their own models

© Manifold Markets, Inc.•Terms•Privacy