SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2605.05802 · 2026

Selective Rollout: Mid-Trajectory Termination for Multi-Sample Agent RL

Xin Wang, Zhiyuan Zhai

arXiv · PDF · Open in the Atlas

Code that ran

13 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 10 failures of our instrument; 0 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/2605.05802.json · badge: /api/ran/2605.05802.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
zhiyuanZhai20/selective-rollout canonical (classifier label)
classifier basis: census
13 ran · 0 no recorded run
FunctionStatusWhere it lives
divergence_at_K Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/divergence.py
pointer only (licence: NONE) · get_code("d17f3c5d626195d7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

followup_user_message Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/prompts.py
pointer only (licence: NONE) · get_code("0c516a2e6bb36237")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

group_reward_stats Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/divergence.py
pointer only (licence: NONE) · get_code("5b8db30b398f2e3f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

group_reward_variance Ran
ran, violated the contract we drafted
zhiyuanZhai20/selective-rollout/src/divergence.py
pointer only (licence: NONE) · get_code("5819c3b6f199f733")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

initial_user_message Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/prompts.py
pointer only (licence: NONE) · get_code("4d7355c6280bd225")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

list_games Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/env.py
pointer only (licence: NONE) · get_code("c7553a721db08d21")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

parse_action Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/prompts.py
pointer only (licence: NONE) · get_code("17942fd27330ceb9")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

pop Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/queue_fs.py
pointer only (licence: NONE) · get_code("2cacdba9dec0b207")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

remaining Ran
ran, honoured the contract we drafted
zhiyuanZhai20/selective-rollout/src/queue_fs.py
pointer only (licence: NONE) · get_code("13a91532a4ad9bc2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

safe_savefig Ran
ran, honoured the contract we drafted
zhiyuanZhai20/selective-rollout/src/figutil.py
pointer only (licence: NONE) · get_code("35c1f7ccae9126e3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

spearman_by_task_type Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/analysis.py
pointer only (licence: NONE) · get_code("683c94d9c4222b3e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

spearman_table Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zhiyuanZhai20/selective-rollout/src/analysis.py
pointer only (licence: NONE) · get_code("5609356339d9cacb")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

verdict Ran
ran; our contract draft was wrong, not the code
zhiyuanZhai20/selective-rollout/src/analysis.py
pointer only (licence: NONE) · get_code("5fc62a7556363e8d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zhiyuanZhai20/selective-rollout when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 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 13 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

Group-relative RL training (GRPO) samples a small group of parallel rollouts for every training prompt and uses their within-group reward spread to compute per-trajectory advantages. In agentic environments each rollout is a long multi-turn dialogue with one LLM call per step, so this multi-sample multiplier dominates the total training cost. When every rollout of a prompt ends with the same reward, the group has zero reward variance and contributes no gradient, so the extra rollouts add no information; such groups are common in practice (typically around 40% of all groups), so the wasted-compute fraction is substantial rather than marginal. Existing methods filter such groups at the prompt level, either after their rollouts are paid for or before any rollout begins, but both decide without using information that becomes available during the rollout itself. We instead ask whether the in-group divergence between the partial trajectories at an intermediate step can already predict that the group will be zero-variance: when the parallel rollouts have already converged on the same action prefix, the group is on track to produce a single reward, and we can stop early. We propose a one-parameter gate that stops a group when the mean pairwise prefix edit distance between its partial action sequences falls below a threshold. On a 60-iteration on-policy GRPO run on ALFWorld with Qwen2.5-7B, averaged over four random seeds, the gated arm finishes 10.7% faster in wall-clock (bootstrap 95% CI excludes 0) and shifts held-out success rate on 50 unseen tasks by +2.5 pp, with the held-out gain tracing to a measurable reduction in zero-advantage gradient-batch dilution. Code: https://github.com/zhiyuanZhai20/selective-rollout.

For agents

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

get_harvested_code_for_paper("2605.05802")
get_code_for_paper("2605.05802")
have("2605.05802")

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

curl https://syntology.ai/api/ran/2605.05802.json

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

[![Syntology run record](https://syntology.ai/api/ran/2605.05802.svg)](https://syntology.ai/paper/2605.05802)

Connect an agent — have() is free.