Papers › Reinforcement Learning Teachers of Test Time Scaling

Reinforcement Learning Teachers of Test Time Scaling

10 Jun 2025arXiv:2506.08388archive 2025-07-28

Edoardo Cetin, Tianyu Zhao, Yujin Tang

Training reasoning language models (LMs) with reinforcement learning (RL) for one-hot correctness inherently relies on the LM being able to explore and solve its task with some chance at initialization. Furthermore, a key use case of reasoning LMs is to act as teachers for distilling new students and cold-starting future RL iterations rather than being deployed themselves. From these considerations, we introduce a new framework that avoids RL's exploration challenge by training a new class of Reinforcement-Learned Teachers (RLTs) focused on yielding the most effective downstream distillation. RLTs are prompted with both the question and solution to each problem, and tasked to simply "connect-the-dots" with detailed explanations tailored for their students. We train RLTs with dense rewards obtained by feeding each explanation to the student and testing its understanding of the problem's solution. In practice, the raw outputs of a 7B RLT provide higher final performance on competition and graduate-level tasks than existing distillation and cold-starting pipelines that collect and postprocess the reasoning traces of orders of magnitude larger LMs. Furthermore, RLTs maintain their effectiveness when training larger students and when applied zero-shot to out-of-distribution tasks, unlocking new levels of efficiency and re-usability for the RL reasoning framework.

PaperPDFCode 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 25 code samples harvested from 1 repository linked to this paper: 18 ran (of which 0 constructed an object rather than computing a result; 18 with no instrument failure: 0 honoured, 0 violated, 18 with no contract checked; 0 where Syntology's instrument failed) · 7 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: found in paper text by Syntology: 25 samples from 1 repository, 18 ran (18 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 on 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

SakanaAI/RLT found in paper text by SyntologyApache-2.0Syntology: 25 samples, 18 ran (0 instrument-failed) · 7 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: 18 ran (of which 0 constructed an object rather than computing a result; 18 with no instrument failure: 0 honoured, 0 violated, 18 with no contract checked; 0 where Syntology's instrument failed) · 7 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.

18ran
7unverified

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 SakanaAI/RLT. “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.

SakanaAI/RLT found in paper text by Syntology 25 samples, 18 ran (0 instrument-failed) · 7 unverified
add_indices custom_data/sft_data.py ran Apache-2.0 (permissive) · 9d2579ff21e86b0f · report fixture: signature ds:Dataset/0/? · returned Dataset · level recorded 2026-09-25
combine_items trainers/teacher_rewards.py ran Apache-2.0 (permissive) · 57900181a4f0bebc · report fixture: signature items:list/1/Tensor · returned Tensor · level recorded 2026-09-25
combine_list_elements trainers/teacher_rewards.py ran Apache-2.0 (permissive) · c0c85c73c543f13d · report fixture: signature list_of_lists:list/1/list · returned list · level recorded 2026-09-25
extract_until custom_data/reasoning_datasets_info.py ran fingerprinted Apache-2.0 (permissive) · ad03b9361505c6ba · report fixture: signature s1:str/0/str, s2:str/0/str · returned str · level recorded 2026-09-25
find_valid_subsequence trainers/teacher_base.py ran fingerprinted Apache-2.0 (permissive) · 213cd812e71a60a6 · report fixture: signature sub:list/1/int, seq:list/1/int · returned list · level recorded 2026-09-25
fix_pad_token hydra_utils.py ran Apache-2.0 (permissive) · d2cba249caa4b1ca · report fixture: signature tokenizer:MockTokenizer/0/?, model_name:str/0/str, unsafe:bool/0/bool · returned MockTokenizer · level recorded 2026-09-25
get_fsdp_wrap_policy trainers/utils_fsdp.py ran Apache-2.0 (permissive) · 411e1acfc478d4c8 · report fixture: signature module:callable/0/?, config:dict/0/?, is_lora:bool/0/bool · returned partial · level recorded 2026-09-25
get_special_token_values custom_data/utils.py ran Apache-2.0 (permissive) · 8ef0bcf6290a2cb8 · report fixture: signature tokenizer:MockTokenizer/0/?, model_name:str/0/str · returned tuple · level recorded 2026-09-25
get_tags custom_data/reasoning_datasets_info.py ran Apache-2.0 (permissive) · 1125a84e5ed38e6c · report fixture: signature tag:str/0/str, format:str/0/str · returned tuple · level recorded 2026-09-25
get_top_dataset trainers/data_reward_scorer.py ran Apache-2.0 (permissive) · d4b4f689be74f50e · report fixture: signature dataset:SimpleDataset/0/?, score_column:str/0/str, number_to_keep:int/0/int · returned Subset · level recorded 2026-09-25
get_top_hf_dataset trainers/data_reward_scorer.py ran Apache-2.0 (permissive) · af28ce60656bad14 · report fixture: signature dataset:MockDataset/0/?, score_column:str/0/str, number_to_keep:int/0/int, reverse:bool/0/bool · returned MockDataset · level recorded 2026-09-25
grab_text_between custom_data/reasoning_datasets_info.py ran fingerprinted Apache-2.0 (permissive) · 42c4a4c44667e48a · report fixture: signature s:str/0/str, start:str/0/str, end:str/0/str · returned str · level recorded 2026-09-25
instantiate_from_target trainers/data_reward_scorer.py ran Apache-2.0 (permissive) · 1d1f8918c9a7700d · report fixture: signature cfg:DictConfig/0/?, kwargs:dict/0/? · returned dict · level recorded 2026-09-25
is_subsequence trainers/teacher_base.py ran Apache-2.0 (permissive) · 726af6b132a154e6 · report fixture: signature sub:list/1/int, seq:list/1/int · returned bool · level recorded 2026-09-25
is_tensor in 9 papers trainers/teacher_base.py ran fingerprinted Apache-2.0 (permissive) · 090ebd62480b897f · report fixture: signature t:Tensor/2/float · returned bool · level recorded 2026-09-25
override_system_prompt custom_data/utils.py ran Apache-2.0 (permissive) · 41b364280acc70df · report fixture: signature tokenizer_or_tokenizer_name:callable/0/?, new_system_prompt:str/0/str · returned MockTokenizer · level recorded 2026-09-25
selective_log_softmax trainers/grpo.py ran Apache-2.0 (permissive) · 39cc9bf845bc3e8d · report fixture: signature logits:Tensor/3/float, index:Tensor/2/int · returned Tensor · level recorded 2026-09-25
to_torch_tensor trainers/teacher_rewards.py ran Apache-2.0 (permissive) · 9c58d17145ebc986 · report fixture: signature data:Tensor/2/float, device:str/0/str, dtype:dtype/0/? · returned Tensor · level recorded 2026-09-25
get_init_weight_context_manager trainers/utils_fsdp.py unverified Apache-2.0 (permissive) · 00ce1c00f50d9695 · report
get_process_line_fn custom_data/sft_data.py unverified Apache-2.0 (permissive) · 901e9003ba5b945f · report
get_process_line_fn custom_data/teacher_data.py unverified Apache-2.0 (permissive) · 482691d35524f06f · report
get_teacher_data_mask_delimiter custom_data/teacher_data.py unverified Apache-2.0 (permissive) · 5fe3db6bb630006c · report
init_fn trainers/utils_fsdp.py unverified Apache-2.0 (permissive) · 376e45ad277a20e3 · report
load_formatted_sft_dataset custom_data/sft_data.py unverified Apache-2.0 (permissive) · 2129b78fe0290670 · report
process_assistant_message custom_data/teacher_data.py unverified Apache-2.0 (permissive) · 2a27f4d8ec953177 · report

Tasks

Reinforcement LearningReinforcement Learning (RL)reinforcement-learning

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