Papers › Train Once, Get a Family: State-Adaptive Balances for Offline-to-Online Reinforcement Learning

Train Once, Get a Family: State-Adaptive Balances for Offline-to-Online Reinforcement Learning

27 Oct 2023NeurIPS 2023 11arXiv:2310.17966archive 2025-07-28

Shenzhi Wang, Qisen Yang, Jiawei Gao, Matthieu Gaetan Lin, Hao Chen, Liwei Wu, Ning Jia, Shiji Song, Gao Huang

Offline-to-online reinforcement learning (RL) is a training paradigm that combines pre-training on a pre-collected dataset with fine-tuning in an online environment. However, the incorporation of online fine-tuning can intensify the well-known distributional shift problem. Existing solutions tackle this problem by imposing a policy constraint on the policy improvement objective in both offline and online learning. They typically advocate a single balance between policy improvement and constraints across diverse data collections. This one-size-fits-all manner may not optimally leverage each collected sample due to the significant variation in data quality across different states. To this end, we introduce Family Offline-to-Online RL (FamO2O), a simple yet effective framework that empowers existing algorithms to determine state-adaptive improvement-constraint balances. FamO2O utilizes a universal model to train a family of policies with different improvement/constraint intensities, and a balance model to select a suitable policy for each state. Theoretically, we prove that state-adaptive balances are necessary for achieving a higher policy performance upper bound. Empirically, extensive experiments show that FamO2O offers a statistically significant improvement over various existing methods, achieving state-of-the-art performance on the D4RL benchmark. Codes are available at https://github.com/LeapLabTHU/FamO2O.

PaperPDFConference PDFCodeCode Syntology ranMethod as code

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 25 code samples harvested from 1 repository linked to this paper: 17 ran (of which 5 constructed an object rather than computing a result; 12 with no instrument failure: 1 honoured, 0 violated, 11 with no contract checked; 5 where Syntology's instrument failed) · 8 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; 3 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 25 samples from 1 repository, 17 ran (12 with no instrument failure; 5 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 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 1 of the 17 samples that ran has no recorded date.

leaplabthu/famo2o officialmentioned in paperjaxMITSyntology: 25 samples, 17 ran (5 instrument-failed) · 8 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

25 samples harvested: 17 ran (of which 5 constructed an object rather than computing a result; 12 with no instrument failure: 1 honoured, 0 violated, 11 with no contract checked; 5 where Syntology's instrument failed) · 8 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
2ran · our draft was wrong
3ran; the contract check could not run
11ran
8unverified

Licence: 0 of the 25 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from leaplabthu/famo2o. “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. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

leaplabthu/famo2o official repository 25 samples, 17 ran (5 instrument-failed) · 8 unverified
flatten in 178 papers
census: 4 implementations from 7 papers, 4 distinct outputs across 4 buckets, one shared input per bucket
jax_cql/viskit/core.py ran · honoured contract MIT (permissive) · 20922c79a0929bbf · report fixture: signature l:list/1/list · returned list · level recorded, date not recorded
Critic in 60 papers jax_iql/family_learner.py ran MIT (permissive) · 099783f7940c8b90 · report fixture: returned Critic · level recorded 2026-09-22
DoubleCritic jax_iql/family_learner.py ran MIT (permissive) · da28baee3154ec4c · report fixture: returned DoubleCritic · level recorded 2026-09-22
MLP in 375 papers jax_iql/family_learner.py ran MIT (permissive) · 9d501615317e5db0 · report fixture: returned MLP · level recorded 2026-09-22
Temperature jax_iql/family_learner.py ran MIT (permissive) · ed3cce240e728d5a · report fixture: returned Temperature · level recorded 2026-09-22
ValueCritic jax_iql/family_learner.py ran MIT (permissive) · c12affb13cb328e9 · report fixture: returned ValueCritic · level recorded 2026-09-22
_sample_actions jax_iql/family_learner.py ran; the contract check could not run (RAISES) MIT (permissive) · 92a63ba161798ebe · report fixture: signature rng:ArrayImpl/1/int, actor_def:callable/0/?, actor_params:FrozenDict/0/?, observations:ndarray/2/float, temperature:float/0/float, deterministic:bool/0/bool · returned tuple · level recorded 2026-09-22
default_init in 38 papers jax_iql/family_learner.py ran · our draft was wrong MIT (permissive) · 29e1a06c6560973e · report fixture: signature scale:ArrayImpl/0/float · returned function · level recorded 2026-09-22
default_init in 38 papers jax_iql/common.py ran MIT (permissive) · 718462716d78ea99 · report fixture: signature scale:ArrayImpl/0/float · returned function · level recorded 2026-09-24
encode in 73 papers jax_cql/cql_finetune/conservative_sac_finetune_main.py ran MIT (permissive) · 1689a9bc9e862217 · report fixture: signature name:str/0/str, length:int/0/int · returned str · level recorded 2026-09-24
load_progress in 7 papers jax_cql/viskit/core.py ran MIT (permissive) · 965e75a1245dd833 · report fixture: signature progress_csv_path:str/0/str · returned dict · level recorded 2026-09-24
multiple_action_q_function in 6 papers jax_cql/JaxCQL/model.py ran MIT (permissive) · 18c566cc009ddb07 · report fixture: signature forward:callable/0/? · returned function · level recorded 2026-09-24
sample_actions jax_iql/family_learner.py ran; the contract check could not run (RAISES) MIT (permissive) · b846a1c39ddf42fe · report fixture: signature rng:ArrayImpl/1/int, actor_def:callable/0/?, actor_params:dict/0/?, observations:ndarray/2/float, temperature:float/0/float, deterministic:bool/0/bool · returned tuple · level recorded 2026-09-22
sin_cos_skill_func jax_iql/family_learner.py ran; the contract check could not run (RAISES) fingerprinted MIT (permissive) · e471b8d548b33cb3 · report fixture: signature beta_array:ndarray/1/float, n:float/0/float, d:int/0/int · returned ArrayImpl · level recorded 2026-09-22
sin_cos_skill_func jax_cql/cql_finetune/family_utils.py ran MIT (permissive) · a81da1617be5131d · report fixture: signature beta_array:ndarray/1/float, n:float/0/float, d:int/0/int · returned ndarray · level recorded 2026-09-24
unique in 30 papers jax_cql/viskit/core.py ran · our draft was wrong fingerprinted MIT (permissive) · 6c7a4a9349a395e9 · report fixture: signature l:list/1/int · returned list · level recorded 2026-09-22
update_target_network in 5 papers jax_cql/JaxCQL/model.py ran MIT (permissive) · f23c0c9faf00d830 · report fixture: signature main_params:dict/0/?, target_params:dict/0/?, tau:float/0/float · returned dict · level recorded 2026-09-26
FamilyLearner jax_iql/family_learner.py unverified MIT (permissive) · be0aac11e3445957 · report
Model in 227 papers jax_iql/family_learner.py unverified MIT (permissive) · bde48e098e1d0bb3 · report
NormalTanhPolicy jax_iql/family_learner.py unverified MIT (permissive) · a441b8020777a636 · report
_family_update_jit jax_iql/family_learner.py unverified MIT (permissive) · 9fee1b402cdb18a3 · report
target_update in 9 papers jax_iql/family_learner.py unverified MIT (permissive) · 958fa187e9bfa7b1 · report
update in 31 papers jax_iql/family_learner.py unverified MIT (permissive) · 9645d2497b243871 · report
update_q jax_iql/family_learner.py unverified MIT (permissive) · 5739da32d5d6bf96 · report
update_v jax_iql/family_learner.py unverified MIT (permissive) · 3ba42b0ed87afaa0 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it.

1 assembly on this paper: 1 harvested, 0 generated.

harvested Assembly for Family Offline-to-Online RL (FamO2O) (graph method: Family Offline-to-Online RL (FamO2O)). 18 bricks; individually on a fixture: 10 ran (of which 5 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 1 violated, 5 with no contract checked; 4 where Syntology's instrument failed) · 8 unverified. The assembly has not been run end-to-end.

  1. FamilyLearner entry · level 0 · unverified · be0aac11e3445957
  2. DoubleCritic helper · level 1 · ran · da28baee3154ec4c
  3. Model helper · level 0 · unverified · bde48e098e1d0bb3
  4. NormalTanhPolicy helper · level 0 · unverified · a441b8020777a636
  5. Temperature helper · level 1 · ran · ed3cce240e728d5a
  6. ValueCritic helper · level 1 · ran · c12affb13cb328e9
  7. _family_update_jit update_rule · level 0 · unverified · 9fee1b402cdb18a3
  8. sample_actions helper · level 1 · ran; the contract check could not run (RAISES) · b846a1c39ddf42fe
  9. sin_cos_skill_func helper · level 1 · ran; the contract check could not run (RAISES) · e471b8d548b33cb3
  10. Critic helper · level 1 · ran · 099783f7940c8b90
  11. MLP helper · level 1 · ran · 9d501615317e5db0
  12. default_init helper · level 1 · ran · our draft was wrong · 29e1a06c6560973e
  13. target_update helper · level 0 · unverified · 958fa187e9bfa7b1
  14. update helper · level 0 · unverified · 9645d2497b243871
  15. update_q helper · level 0 · unverified · 5739da32d5d6bf96
  16. update_v helper · level 0 · unverified · 3ba42b0ed87afaa0
  17. _sample_actions helper · level 1 · ran; the contract check could not run (RAISES) · 92a63ba161798ebe
  18. loss helper · level 1 · ran · violated contract · 13df74f8d96dca61

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2310.17966") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

D4RLReinforcement Learning (RL)

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

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