REVIEW 3 cited by
SELF-[IN]CORRECT: LLMs Struggle with Discriminating Self-Generated Responses
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Can LLMs consistently improve their previous outputs for better results? For this to be true, LLMs would need to be better at discriminating among previously-generated alternatives, than generating initial responses. We explore the validity of this hypothesis in practice. We first formulate a unified framework that allows us to compare the generative and discriminative capability of any model on any task. In our resulting experimental analysis of several open-source and industrial LLMs, we observe that models are not reliably better at discriminating among previously-generated alternatives than generating initial responses. This finding challenges the notion that LLMs may be able to enhance their performance only through their own judgment.
Forward citations
Cited by 3 Pith papers
-
Training Language Model to Critique for Better Refinement
RCO trains critic LLMs using refinement utility, the extent to which a critique improves a revised answer, as the reward, outperforming direct critique preference methods across five tasks.
-
Judging Is Not Enumerating: Silent Omissions in LLM-Authored Acceptable Sets
LLMs under one-shot greedy decoding enumerate or materialize acceptable sets much worse than they judge membership, a gap that persists across scale, family, and generation and is dominated by omissions.
-
Unlocking Recursive Thinking of LLMs: Alignment via Refinement
An offline alignment pipeline using reward-filtered self-refinement data and long chain-of-thought SFT raises an 8B model's AlpacaEval 2 win rate from 25.0% to 51.0% with roughly 14k training examples.
Discussion (0). Continue with ORCID to comment.