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

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2503.18929")

Code

Syntology Ran 0 of 10 code samples harvested from 1 repository linked to this paper; 10 have no recorded run.

By repository: community (archive-listed): 10 samples from 1 repository, 0 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

bbartoldson/TBA mentioned on GitHubpytorchApache-2.0 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; 0 ran; 0 honoured the contract we drafted; 10 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

10unverified

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. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (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 from a repository marked official are listed first. 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 four 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; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). 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.

build_example_dataloader bbartoldson/TBA/src/dist_data_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · d0b3a48ae72ccfc4 · report
example_prepare_dataset bbartoldson/TBA/src/dist_data_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 2ed37256b49d8561 · report
extract_prediction bbartoldson/TBA/src/gsm8k_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · a26a4bc2f82b9ec2 · report
format_and_tokenize bbartoldson/TBA/src/gsm8k_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 150dea51d807fbaa · report
ignore_prompt_labels bbartoldson/TBA/src/kl.py community (archive-listed) unverified Apache-2.0 (permissive) · d70961db4cf293ba · report
next_free_port bbartoldson/TBA/src/dist_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 5503c179cc4b29b0 · report
parse_number bbartoldson/TBA/src/gsm8k_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · cd83783cc6325cf1 · report
prepare_dataset bbartoldson/TBA/tba_tldr.py community (archive-listed) unverified Apache-2.0 (permissive) · a2e246bb45f45630 · report
prepare_deepspeed bbartoldson/TBA/src/utils.py community (archive-listed) unverified Apache-2.0 (permissive) · e0030346eb43e3a2 · report
split_dataset_indices bbartoldson/TBA/src/dist_data_utils.py community (archive-listed) unverified Apache-2.0 (permissive) · 9860770ecc9b4ebe · 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