HFEPX · Eval paper review
Yunsheng Lu, Zijiang Yang, Licheng Pan, Zhixuan Chu
Published
Apr 15, 2026
Citations
0
Trust level
Provisional
Usefulness score
Unavailable
Extraction confidence
0% (Provisional)
Derived from abstract and metadata only.
Signals refreshed
Apr 16, 2026
Signal extraction is still processing. This page currently shows metadata-first guidance until structured protocol fields are ready.
This page is a lightweight research summary built from the abstract and metadata while deeper extraction catches up.
All signals on this page are inferred from the abstract only and may be inaccurate. Do not use this page as a primary protocol reference.
Best use
Background context only
Use if you need
A provisional background reference while structured extraction finishes.
What to verify
Read the full paper before copying any benchmark, metric, or protocol choices.
Main weakness
This page is still relying on abstract and metadata signals, not a fuller protocol read.
Eval-fit score is unavailable until extraction completes.
If you are doing eval pipeline work, start here
Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone. Most prior work weakens these cues directly by penalizing or controlling specific artifacts, but it does not explicitly encourage the model to ground preferences in the prompt's intent. We learn a decoder that maps a candidate answer to the latent intent embedding of the input. The reconstruction error is used as a signal to regularize the reward model training. We provide theoretical evidence that this signal emphasizes prompt-dependent information while suppressing prompt-independent shortcuts. Across math, helpfulness, and safety benchmarks, the decoder selects shorter and less sycophantic candidates with 0.877 accuracy. Incorporating this signal into RM training in Gemma-2-2B-it and Gemma-2-9B-it increases RewardBench accuracy from 0.832 to 0.868. For Best-of-N selection, our framework increases length-controlled win rates while producing shorter outputs, and remains robust to lengthening and mild off-topic drift in controlled rewrite tests.
These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.
None explicit
No explicit feedback protocol extracted.
"Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone."
Automatic metrics
Includes extracted eval setup.
"Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone."
Not reported
No explicit QC controls found.
"Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone."
MATH
Useful for quick benchmark comparison.
"Across math, helpfulness, and safety benchmarks, the decoder selects shorter and less sycophantic candidates with 0.877 accuracy."
Accuracy
Useful for evaluation criteria comparison.
"Across math, helpfulness, and safety benchmarks, the decoder selects shorter and less sycophantic candidates with 0.877 accuracy."
Unknown
Rater source not explicitly reported.
"Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone."
This page is using abstract-level cues only right now. Treat the signals below as provisional.
Evaluation fields are inferred from the abstract only.
Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone.
Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.
