SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.09760 · ICLR · 2025

Bootstrapping Language Models with DPO Implicit Rewards

Tianyu Pang, Qian Liu, Min Lin, Zichen Liu, Chao Du, Changyu Chen, Arunesh Sinha, Pradeep Varakantham

arXiv · PDF · Open in the Atlas

Code that ran

None of the 2 samples (functions and classes) lifted from this paper's linked repositories has a recorded run on a synthesized input (2 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2406.09760.json · badge: /api/ran/2406.09760.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
sail-sg/dice canonical (classifier label)
classifier basis: classifier_2025
0 ran · 2 no recorded run
FunctionStatusWhere it lives
exact_match_numbers Not yet run
unverified (harvested, no recorded run)
sail-sg/dice/self_alignment/utils/eval_util.py
code served (permissive licence) · get_code("238bba8d62db6b40")
How it ran

no invocation recorded. this sample carries JSON test cases instead of a setup block; get_reference_implementation serves them with the level they back.

extract_numbers Not yet run
unverified (harvested, no recorded run)
sail-sg/dice/self_alignment/utils/eval_util.py
code served (permissive licence) · get_code("d9accffd5c3b9c10")
How it ran

no invocation recorded. this sample carries JSON test cases instead of a setup block; get_reference_implementation serves them with the level they back.

Also present in other papers' code

Of 2 samples lifted from this paper's linked repositories, 2 carry a function identity and 0 do not. Of the 2 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 2 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Human alignment in large language models (LLMs) is an active area of research. A recent groundbreaking work, direct preference optimization (DPO), has greatly simplified the process from past work in reinforcement learning from human feedback (RLHF) by bypassing the reward learning stage in RLHF. DPO, after training, provides an implicit reward model. In this work, we make a novel observation that this implicit reward model can by itself be used in a bootstrapping fashion to further align the LLM. Our approach is to use the rewards from a current LLM to construct a preference dataset, which is then used in subsequent DPO rounds. We incorporate two refinements to further improve our approach: 1) length-regularized reward shaping to make the preference dataset length-unbiased; 2) experience replay to enhance the quality of the preference dataset. Our approach, named self-alignment with DPO ImpliCit rEwards (DICE), shows great improvements in alignment. It achieves an increase of more than 8% in lengthcontrolled win rate on AlpacaEval 2 for all the different base models that we tried, without relying on external feedback. Our code is available at https: //github.com/sail-sg/dice. * Equal contribution. The project was done during Changyu Chen's internship at Sea AI Lab.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2406.09760")
get_code_for_paper("2406.09760")
have("2406.09760")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2406.09760.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2406.09760.svg)](https://syntology.ai/paper/2406.09760)

Connect an agent — have() is free.