Will DPO variants mostly replace RLHF before EOY 2024?
Basic
11
Ṁ410Dec 31
40%
chance
1D
1W
1M
ALL
Resolves YES if ≥3 out of the 5 top LLMs on Chabot Arena use (a variant of) DPO on 2024/12/31.
This question is managed and resolved by Manifold.
Get
1,000
and3.00
Sort by:
I think this market fundamentally cannot resolve as stated because Chatbot Arena includes closed LLMs which do not disclose the details of how they have been finetuned. GPT-4 and Bard might already be using DPO and we wouldn't know about it. To my chagrin, some people (for example Nous Research) categorize DPO as a variant of RLHF, so there's plausible deniability whenever OpenAI or Google refer to their finetuning as "RLHF"
@NoraBelrose that's a good point. If there's uncertainty about close source models at the end of the year I might resolve N/A or push the deadline in the hope that the information will surface later.
Related questions
Related questions
Will DPO or an Explicitly DPO-based Technique be Used to Train a Public Frontier Lab LLM Before Jan 1 2025?
78% chance
Who will die by EOY 2024?
Will a single model have all the upsides o1-style RL with none of the downsides at 2027?
54% chance
Will the ωB97M-V functional (DFT) be widely regarded as obsolete by EOY 2027?
50% chance
Will I still never get COVID though EOY 2025?
59% chance
Who will die by EOY 2025
Will I get follistatin gene therapy by EOY 2026?
40% chance
Will Rung 6 density functionals be widespread by EOY 2025?
36% chance
What will Manifolders mostly use LLMs for, by EOY 2025?
Will the main LK99 market ever reach 99% EOY 2024
4% chance