00:00:00

Share Your Feedback 🏝️

Why is Your Language Model a Poor Implicit Reward Model?

Why is Your Language Model a Poor Implicit Reward Model?

MinWoo(Daniel) Park | Tech Blog

Read more
Previous: How Well Does GPT-4o Understand Vision Next: SAS

Why is Your Language Model a Poor Implicit Reward Model?

  • Related Project: Private
  • Category: Paper Review
  • Date: 2025-07-11

Why is Your Language Model a Poor Implicit Reward Model?

  • url https://arxiv.org/abs/2507.07981
  • pdf https://arxiv.org/pdf/2507.07981
  • abstract Reward models are key to language model post-training and inference pipelines. Conveniently, recent work showed that every language model defines an implicit reward model (IM-RM), without requiring any architectural changes. However, such IM-RMs tend to generalize worse, especially out-of-distribution, compared to explicit reward models (EX-RMs) that apply a dedicated linear head over the hidden representations of a language model. The existence of a generalization gap is puzzling, as EX-RMs and IM-RMs are nearly identical. They can be trained using the same data, loss function, and language model, and differ only in how the reward is computed. Towards a fundamental understanding of the implicit biases underlying different reward model types, we investigate the root cause of this gap. Our main finding, backed by theory and experiments, is that IM-RMs rely more heavily on superficial token-level cues. Consequently, they often generalize worse than EX-RMs under token-level distribution shifts, as well as in-distribution. Furthermore, we provide evidence against alternative hypotheses for the generalization gap. Most notably, we challenge the intuitive claim that IM-RMs struggle in tasks where generation is harder than verification because they can operate both as a verifier and a generator. Taken together, our results highlight that seemingly minor design choices can substantially impact the generalization behavior of reward models.
Previous: How Well Does GPT-4o Understand Vision Next: SAS

post contain ""

    No matching posts found containing ""