RLHF in LLMs

The cluster focuses on discussions attributing specific behaviors, personalities, and response styles in large language models to Reinforcement Learning from Human Feedback (RLHF), distinguishing it from base model pre-training.

➡️ Stable 0.9x AI & Machine Learning
3,102
Comments
10
Years Active
5
Top Authors
#8076
Topic ID

Activity Over Time

2017
1
2018
2
2019
7
2020
7
2021
7
2022
115
2023
1,195
2024
642
2025
1,065
2026
61

Keywords

CTGT LLM openai.com RLFH C.A P.I AI HN RL RLMF model tuning feedback reinforcement instruction training llms tuned fine tuning models

Sample Comments

drdeca May 21, 2023 View on HN

Are you taking the RLHF into account when you say so?

RLHF is probably the reason for this.

VivaLaPanda Feb 26, 2024 View on HN

It's almost certainly the RLHF, not the base model.

whimsicalism Apr 12, 2023 View on HN

Read about RLHF, i think you are misunderstanding what this will be used for.

piperswe Jun 5, 2025 View on HN

I think it's part of the RLHF tuning as well

zarzavat Mar 14, 2023 View on HN

He’s talking about RLHF - reinforcement learning with human feedback (the process that trained ChatGPT), the training data for which is not publicly available.And the point is that you don’t need to RLHF as long as you have access to another model that has been trained with RLHF that you can blackbox.

7moritz7 Nov 10, 2025 View on HN

Hasn't RLHF and with LLM feedback been around for years now

visarga May 12, 2023 View on HN

RLHF your AI until you like the output?

sweezyjeezy Sep 5, 2023 View on HN

That's probably more because of RLHF though, they've optimised for certain kind of responses rather than simple model loss on internet text.

drsim Feb 26, 2025 View on HN

It is RLHF if I understand correctly.