[HN Gopher] Binary Retrieval-Augmented Reward Mitigates Hallucin...
___________________________________________________________________
Binary Retrieval-Augmented Reward Mitigates Hallucinations
Author : MarlonPro
Score : 32 points
Date : 2025-10-21 16:14 UTC (6 hours ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| amflare wrote:
| > Existing mitigation approaches often degrade performance on
| open-ended generation and downstream tasks, limiting their
| practical utility. [...] Unlike continuous reward schemes, our
| approach assigns a reward of one only when the model's output is
| entirely factually correct, and zero otherwise.
|
| Someone correct me if I am wrong, as I'm am on the very edge of
| this space looking in, but does this mean that they are using a
| "degraded performance with fewer hallucinations" model to fact
| check the "more powerful yet prone to hallucinations" model?
| svnt wrote:
| Also on the edge, but it appears they are relying on the
| search-augmented identification of conflicts in the generated
| statement, which is an easier task than constructing an answer
| to the question. It also encourages abstention because there
| are no conflicts in "I don't know" (so "mitigating
| hallucinations" and "answering more questions correctly" are
| not necessarily the same thing)
| mNovak wrote:
| My understanding is no, they are collecting a cache of
| documents from the training set, then after pre-training prompt
| about those topics. A separate verifier is given both the
| relevant source documents and generated response, and tasked
| with checking for conflicts in factuality.
|
| They describe using Qwen 32B as the verifier, and the model
| under training is Qwen 8B. So in fact the verifier is beefier
| than the trainee model, though it's unclear if that has to be
| the case as you scale up.
___________________________________________________________________
(page generated 2025-10-21 23:01 UTC)