Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Reaching Beyond the Mode: RL for Distributional Reasoning in Language Models

About

Given a question, a language model (LM) implicitly encodes a distribution over possible answers. In practice, post-training procedures for LMs often collapse this distribution onto a single dominant mode. While this is generally not a problem for benchmark-style evaluations that assume one correct answer, many real-world tasks inherently involve multiple valid answers or irreducible uncertainty. Examples include medical diagnosis, ambiguous question answering, and settings with incomplete information. In these cases, we would like LMs to generate multiple plausible hypotheses, ideally with confidence estimates for each one, and without computationally intensive repeated sampling to generate non-modal answers. This paper describes a multi-answer reinforcement learning approach for training LMs to perform distributional reasoning over multiple answers during inference. We modify the RL objective to enable models to explicitly generate multiple candidate answers in a single forward pass, internalizing aspects of inference-time search into the model's generative process. Across question-answering, medical diagnostic, and coding benchmarks, we observe improved diversity, coverage, and set-level calibration scores compared to single answer trained baselines. Models trained with our approach require fewer tokens to generate multiple answers than competing approaches. On coding tasks, they are also substantially more accurate. These results position multi-answer RL as a principled and compute-efficient alternative to inference-time scaling procedures such as best-of-k. Code and more information can be found at https://multi-answer-rl.github.io/.

Isha Puri, Mehul Damani, Idan Shenfeld, Marzyeh Ghassemi, Jacob Andreas, Yoon Kim• 2026

Related benchmarks

TaskDatasetResultRank
Code GenerationMBPP
Top-1 Acc.49
21
CalibrationMBPP
Top-1 ECE0.37
8
Medical Differential DiagnosesDDXPlus
Avg Correct79
8
Question AnsweringHotPotQA Modified
Average Correctness27
8
CalibrationHotPotQA Hard
Top-1 ECE0.31
8
Chain-of-Thought ReasoningEUREQA (held-out half of hard_5)
Best@321
8
Maze NavigationMaze 100 held-out mazes
Best Success Rate @ 342
8
Function CallingToolRL 80-prompt (held-out)
Best@386.1
8
Multi-hop Question AnsweringMuSiQue 300-question hop-stratified (held-out)
Best@359.9
8
CalibrationDDXPlus
Top-1 ECE0.01
4
Showing 10 of 10 rows

Other info

GitHub

Follow for update