Papers › Group Robust Preference Optimization in Reward-free RLHF

Group Robust Preference Optimization in Reward-free RLHF

30 May 2024arXiv:2405.20304archive 2025-07-28

Shyam Sundhar Ramesh, Yifan Hu, Iason Chaimalas, Viraj Mehta, Pier Giuseppe Sessa, Haitham Bou Ammar, Ilija Bogunovic

Adapting large language models (LLMs) for specific tasks usually involves fine-tuning through reinforcement learning with human feedback (RLHF) on preference data. While these data often come from diverse labelers' groups (e.g., different demographics, ethnicities, company teams, etc.), traditional RLHF approaches adopt a "one-size-fits-all" approach, i.e., they indiscriminately assume and optimize a single preference model, thus not being robust to unique characteristics and needs of the various groups. To address this limitation, we propose a novel Group Robust Preference Optimization (GRPO) method to align LLMs to individual groups' preferences robustly. Our approach builds upon reward-free direct preference optimization methods, but unlike previous approaches, it seeks a robust policy which maximizes the worst-case group performance. To achieve this, GRPO adaptively and sequentially weights the importance of different groups, prioritizing groups with worse cumulative loss. We theoretically study the feasibility of GRPO and analyze its convergence for the log-linear policy class. By fine-tuning LLMs with GRPO using diverse group-based global opinion data, we significantly improved performance for the worst-performing groups, reduced loss imbalances across groups, and improved probability accuracies compared to non-robust baselines.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 20 code samples harvested from 2 repositories linked to this paper: 19 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 1 honoured, 1 violated, 9 with no contract checked; 8 where Syntology's instrument failed) · 1 unverified. Where Syntology's instrument failed: 6 where our contract draft was wrong; 2 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 20 samples from 2 repositories, 19 ran (11 with no instrument failure; 8 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written between 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

rsshyam/Group-robust-preference-optimization officialmentioned in paperpytorchApache-2.0Syntology: 14 samples, 13 ran (7 instrument-failed) · 1 unverified report
rsshyam/Group-robust-preference-optimization-bandits officialmentioned in paperSyntology: 6 samples, 6 ran (1 instrument-failed) · 0 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

20 samples harvested: 19 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 1 honoured, 1 violated, 9 with no contract checked; 8 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
1ran · violated contract
6ran · our draft was wrong
2ran; the contract check could not run
9ran
1unverified

Licence: 7 of the 20 samples are pointer only, meaning Syntology does not serve that copy's text. For 1 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

rsshyam/Group-robust-preference-optimization official repository 14 samples, 13 ran (7 instrument-failed) · 1 unverified
call_chats src/eval/fast_oai.py ran Apache-2.0 (permissive) · 1631c9bb4fb2bec9 · report fixture: signature prompts:list/1/tuple, model:str/0/str, timeout:int/0/int, temperature:float/0/float · returned list · level recorded 2026-09-24
create_filter_dicts plot_scripts/plot_from_wandb_full_metrics.py ran Apache-2.0 (permissive) · b4513951d011854f · report fixture: signature groups:list/1/str, n_epochs:int/0/int, base:bool/0/bool, setting:str/0/str · returned list · level recorded 2026-09-24
create_filter_dicts plot_scripts/plot_from_wandb_paper_plots.py ran Apache-2.0 (permissive) · 33b2f038945518a7 · report fixture: signature groups:list/1/str, n_epochs:int/0/int, base:bool/0/bool, setting:str/0/str · returned list · level recorded 2026-09-24
exponential_moving_average plot_scripts/plot_from_wandb_full_metrics.py ran; the contract check could not run (RAISES) fingerprinted Apache-2.0 (permissive) · bb5e7a9121ff84f3 · report fixture: signature data:ndarray/1/float, alpha:float/0/float · returned ndarray · level recorded 2026-09-24
extract_anthropic_prompt in 16 papers src/preference_datasets.py ran · our draft was wrong Apache-2.0 (permissive) here; the stored copy is pointer only · 12f6803880f9b17e · report fixture: signature prompt_and_response:str/0/str · returned str · level recorded 2026-09-22
get_local_dir in 14 papers src/utils.py ran · our draft was wrong Apache-2.0 (permissive) · 8ab43db002bf984d · report fixture: signature prefixes_to_resolve:list/1/str · returned str · level recorded 2026-09-25
get_local_run_dir in 14 papers src/utils.py ran · our draft was wrong Apache-2.0 (permissive) · 10ab71f27ffd7de8 · report fixture: signature exp_name:str/0/str, local_dirs:list/1/str · returned str · level recorded 2026-09-24
get_remote_file in 16 papers src/utils.py ran Apache-2.0 (permissive) · e7ce4c19888c6df6 · report fixture: signature remote_path:str/0/str, local_path:str/0/str · returned str · level recorded 2026-09-25
get_setting_details plot_scripts/plot_from_wandb_full_metrics.py ran · our draft was wrong Apache-2.0 (permissive) · b6b945a848b730d9 · report fixture: signature setting_key:str/0/str · returned list · level recorded 2026-09-24
get_user_prompt in 5 papers src/eval/win_rate.py ran · our draft was wrong Apache-2.0 (permissive) · 6d492ffbac8f0b24 · report fixture: signature row:dict/0/? · returned str · level recorded 2026-09-24
preference_loss in 9 papers src/loss_utils.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 17df4b678a845f54 · report fixture: signature policy_chosen_logps:Tensor/1/float, policy_rejected_logps:Tensor/1/float, reference_chosen_logps:Tensor/1/float, reference_rejected_logps:Tensor/1/float, beta:float/0/float, label_smoothing:float/0/float, ipo:bool/0/bool, reference_free:bool/0/bool · returned tuple · level recorded 2026-09-24
process_max_fields plot_scripts/visualisations_utils_wandb_api.py ran Apache-2.0 (permissive) · fa80f592f919c3e7 · report fixture: signature runs:list/1/DataFrame, fields:list/1/str, maximum:bool/0/bool, time_field:str/0/str, x_percent_rmv:NoneType/0/? · returned list · level recorded 2026-09-24
process_runs plot_scripts/visualisations_utils_wandb_api.py ran Apache-2.0 (permissive) · 090e5779e4a32ffe · report fixture: signature runs:list/1/DataFrame, field:str/0/str, time_field:str/0/str, agg:str/0/str · returned list · level recorded 2026-09-24
download_runs plot_scripts/visualisations_utils_wandb_api.py unverified Apache-2.0 (permissive) · 3e6c31b9f57871b0 · report
rsshyam/Group-robust-preference-optimization-bandits official repository 6 samples, 6 ran (1 instrument-failed) · 0 unverified
float_list experiments/run_group_linear_bandit_sep_theta_combined_det_ratio_vectorised.py ran · honoured contract no licence file found · pointer only · 8fd2b88127329ba4 · report fixture: signature arg:str/0/str · returned list · level recorded 2026-09-24
get_reward_func experiments/run_group_linear_bandit_sep_theta_combined_det_ratio_vectorised.py ran · our draft was wrong no licence file found · pointer only · 30c4cff9ebf96ddb · report fixture: signature reward_param:ndarray/1/float, feature_func:callable/0/? · returned function · level recorded 2026-09-24
ret_feature_func envs/group_linear_bandit.py ran no licence file found · pointer only · 15990987101d9ed3 · report fixture: signature num_action:int/0/int, state_dim:int/0/int, group_num:int/0/int, feature_type:str/0/str · returned function · level recorded 2026-09-24
ret_feature_func_debug envs/group_linear_bandit.py ran no licence file found · pointer only · a9d43fdf06185cf7 · report fixture: signature num_action:int/0/int, state_dim:int/0/int, group_num:int/0/int · returned function · level recorded 2026-09-24
ret_feature_func_vectorised envs/group_linear_bandit.py ran no licence file found · pointer only · 3a36607e2402e2b7 · report fixture: signature num_action:int/0/int, state_dim:int/0/int, group_num:int/0/int, feature_type:str/0/str · returned function · level recorded 2026-09-24
str_to_bool_list experiments/run_group_linear_bandit_sep_theta_combined_det_ratio_vectorised.py ran · violated contract no licence file found · pointer only · 7b0560da4bbdfc82 · report fixture: signature s:str/0/str · returned list · level recorded 2026-09-24

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

ALIGN

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections