Papers › Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast,...

Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Training

24 Mar 2025arXiv:2503.18929archive 2025-07-28

Brian R. Bartoldson, Siddarth Venkatraman, James Diffenderfer, Moksh Jain, Tal Ben-Nun, Seanie Lee, Minsu Kim, Johan Obando-Ceron, Yoshua Bengio, Bhavya Kailkhura

Reinforcement learning (RL) is a critical component of large language model (LLM) post-training. However, existing on-policy algorithms used for post-training are inherently incompatible with the use of experience replay buffers, which can be populated scalably by distributed off-policy actors to enhance exploration as compute increases. We propose efficiently obtaining this benefit of replay buffers via Trajectory Balance with Asynchrony (TBA), a massively scalable LLM RL system. In contrast to existing approaches, TBA uses a larger fraction of compute on search, constantly generating off-policy data for a central replay buffer. A training node simultaneously samples data from this buffer based on reward or recency to update the policy using Trajectory Balance (TB), a diversity-seeking RL objective introduced for GFlowNets. TBA offers three key advantages: (1) decoupled training and search, speeding up training wall-clock time by 4x or more; (2) improved diversity through large-scale off-policy sampling; and (3) scalable search for sparse reward settings. On mathematical reasoning, preference-tuning, and automated red-teaming (diverse and representative post-training tasks), TBA produces speed and performance improvements over strong 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 10 code samples harvested from 1 repository linked to this paper: 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: community (archive-listed): 10 samples from 1 repository, 7 ran (7 with no instrument failure; 0 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 2026-09-24 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

bbartoldson/TBA mentioned on GitHubpytorchApache-2.0Syntology: 10 samples, 7 ran (0 instrument-failed) · 3 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

10 samples harvested: 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 3 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.

7ran
3unverified

Licence: 0 of the 10 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 bbartoldson/TBA. “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.

bbartoldson/TBA community (archive-listed) 10 samples, 7 ran (0 instrument-failed) · 3 unverified
build_example_dataloader src/dist_data_utils.py ran Apache-2.0 (permissive) · d0b3a48ae72ccfc4 · report fixture: signature dataset:MinimalDataset/0/?, tokenizer:callable/0/?, batch_size:int/0/int · returned DataLoader · level recorded 2026-09-24
extract_prediction src/gsm8k_utils.py ran fingerprinted Apache-2.0 (permissive) · a26a4bc2f82b9ec2 · report fixture: signature text:str/0/str · returned float · level recorded 2026-09-24
format_and_tokenize src/gsm8k_utils.py ran Apache-2.0 (permissive) · 150dea51d807fbaa · report fixture: signature example:dict/0/?, tokenizer:callable/0/? · returned dict · level recorded 2026-09-25
ignore_prompt_labels src/kl.py ran Apache-2.0 (permissive) · d70961db4cf293ba · report fixture: signature batch:dict/0/?, response_token_ids:Tensor/1/int, ignore_index:int/0/int, tokenizer:NoneType/0/? · returned dict · level recorded 2026-09-24
next_free_port in 5 papers src/dist_utils.py ran fingerprinted Apache-2.0 (permissive) · 5503c179cc4b29b0 · report fixture: signature port:int/0/int, max_port:int/0/int · returned int · level recorded 2026-09-25
parse_number src/gsm8k_utils.py ran fingerprinted Apache-2.0 (permissive) · cd83783cc6325cf1 · report fixture: signature text:str/0/str · returned float · level recorded 2026-09-24
split_dataset_indices src/dist_data_utils.py ran fingerprinted Apache-2.0 (permissive) · 9860770ecc9b4ebe · report fixture: signature total_size:int/0/int, K:int/0/int · returned list · level recorded 2026-09-25
example_prepare_dataset src/dist_data_utils.py unverified Apache-2.0 (permissive) · 2ed37256b49d8561 · report
prepare_dataset in 32 papers tba_tldr.py unverified Apache-2.0 (permissive) · a2e246bb45f45630 · report
prepare_deepspeed src/utils.py unverified Apache-2.0 (permissive) · e0030346eb43e3a2 · report

Tasks

DiversityLarge Language ModelMathematical ReasoningRed TeamingReinforcement Learning (RL)

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Experience Replay

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