Self-Rewarding Language Models · HackerPapers discussions

Self-Rewarding Language Models — mentioned by chunhualiao

Sun, 21 Jan 2024 23:10:35 GMT

Quickly scanned https://arxiv.org/abs/2401.10020 . Quite interesting work. The paper's idea is to have a single language model doing both question answering (responding to prompts) and self-evaluating its own answers. Iterative DPO training is used to improve the model's dual capabilities. The authors tried different LLM-as-a-judge promptings to generate a reward score for each answer. A very particular additive 5-point rewarding prompting is found to be the most effective one. The two-step inferencing pipeline (answering questions+evaluating answers) also generates an extra dataset in the…

Paper and discussion history. A mention is not an endorsement.

Self-Rewarding Language Models — mentioned by gfodor

Sun, 21 Jan 2024 04:25:34 GMT

Probably not going to pan out well https://arxiv.org/html/2401.10020v1

Paper and discussion history. A mention is not an endorsement.

Self-Rewarding Language Models — mentioned by djoldman

Sat, 20 Jan 2024 23:53:23 GMT

https://arxiv.org/abs/2401.10020

Paper and discussion history. A mention is not an endorsement.

Self-Rewarding Language Models — HN submission

Fri, 19 Jan 2024 21:24:58 GMT

Self-Rewarding Language Models

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Self-Rewarding Language Models — HN submission

Fri, 19 Jan 2024 03:33:43 GMT

Self-Rewarding Language Models

95 points; 60 discussion replies on HN. Paper and source history. A submission is not an endorsement.