SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.03143 · 2026

Self-Hinting Language Models Enhance Reinforcement Learning

Jiang Bian, Hanze Dong, Christof Monz, Baohao Liao, Xinxing Xu

arXiv · PDF · Open in the Atlas

Code that ran

6 of 10 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; 1 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 4 failures of our instrument; 4 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/2602.03143.json · badge: /api/ran/2602.03143.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
BaohaoLiao/SAGE canonical (classifier label)
classifier basis: plan_row
6 ran · 4 no recorded run
FunctionStatusWhere it lives
extract_boxed Ran
ran; our contract draft was wrong, not the code
BaohaoLiao/SAGE/eval/compute_score.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Andree-9/HiLL, is pointer only (licence: NONE) · get_code("82ab5df795d522c8")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from Andree-9/HiLL, and GitHub detected no licence in that repository 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.

generate_completions Ran
ran; our contract draft was wrong, not the code
BaohaoLiao/SAGE/eval/qwen_evaluation/model_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from fanqiwan/FuseAI, is pointer only (licence: NONE) · get_code("1525d8bc0fc8d5a0")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from fanqiwan/FuseAI, and GitHub detected no licence in that repository 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.

generate_config_from_args Ran
ran; our contract draft was wrong, not the code
BaohaoLiao/SAGE/verl/model_merger/base_model_merger.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("8896f4e2bafe8abc")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from LeapLabTHU/OPDVR, and GitHub detected no licence in that repository 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.

get_dynamic_pipeline_shards Ran
ran, honoured the contract we drafted
BaohaoLiao/SAGE/verl/model_merger/megatron_model_merger.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("94209884dcb2308e")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from LeapLabTHU/OPDVR, and GitHub detected no licence in that repository 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.

normalize_final_answer Ran
ran; our contract draft was wrong, not the code
BaohaoLiao/SAGE/eval/oat_math_grader.py
code served (permissive licence) · get_code("5662e5ee755b541c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

SUBSTITUTIONS = [('an ', ''), ('a ', ''), ('.$', '$'), ('\\$', ''), ('\\ ', ''), (' ', ''), ('mbox', 'text'), (',\\text{and}', ','), ('\\text{and}', ','), ('\\text{m}', '\\text{}')]

REMOVED_EXPRESSIONS = ['square', 'ways', 'integers', 'dollars', 'mph', 'inches', 'ft', 'hours', 'km', 'units', '\\ldots', 'sue', 'points', 'feet', 'minutes', 'digits', 'cents', 'degrees', 'cm', 'gm', 'pounds', 'meters', 'meals', 'edges', 'students', 'childrentickets', 'multiples', '\\text{s}', '\\text{.}', '\\text{\ns}', '\\text{}^2', '\\text{}^3', '\\text{\n}', '\\text{}', '\\mathrm{th}', '^\\circ', '^{\\circ}', '\\;', ',\\!', '{,}', '"', '\\dots']

def normalize_final_answer(final_answer: str) -> str:
    """
    Normalize a final answer to a quantitative reasoning question.
    This code comes from https://arxiv.org/pdf/2206.14858.pdf, page18.
    """
    for before, after in SUBSTITUTIONS:
        final_answer = final_answer.replace(before, after)
    for expr in REMOVED_EXPRESSIONS:
        final_answer = final_answer.replace(expr, '')
    final_answer = re.sub('(.*?)(\\$)(.*?)(\\$)(.*)', '$\\3$', final_answer)
    final_answer = re.sub('(\\\\text\\{)(.*?)(\\})', '\\2', final_answer)
    final_answer = re.sub('(\\\\textbf\\{)(.*?)(\\})', '\\2', final_answer)
    final_answer = re.sub('(\\\\overline\\{)(.*?)(\\})', '\\2', final_answer)
    final_answer = re.sub('(\\\\boxed\\{)(.*)(\\})', '\\2', final_answer)
    final_answer = re.sub('(frac)([^{])(.)', 'frac{\\2}{\\3}', final_answer)
    final_answer = re.sub('(sqrt)([^{])', 'sqrt{\\2}', final_answer)
    final_answer = final_answer.replace('$', '')
    if final_answer.replace(',', '').isdigit():
        final_answer = final_answer.replace(',', '')
    return final_answer

ARGS = {
    'final_answer': '\\boxed{42}'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
repeatness Ran
ran, violated the contract we drafted
BaohaoLiao/SAGE/eval/oat_math_grader.py
code served (permissive licence) · get_code("3dcb40e5ce9bac48")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

# Analyzing the function
# repeatness() takes a single parameter: s (a string)
# The function computes a "repeatness" metric based on suffix arrays and LCP arrays

# Simple test cases
ARGS = {
    "s": "abcabcabc"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_weight_loader Not yet run
unverified (harvested, no recorded run)
BaohaoLiao/SAGE/verl/models/weight_loader_registry.py
code served (permissive licence) · get_code("acf240079d434746")
How it ran

no invocation recorded.

get_weight_saver Not yet run
unverified (harvested, no recorded run)
BaohaoLiao/SAGE/verl/models/weight_loader_registry.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from mit-han-lab/SMEPO, is pointer only (licence: NONE) · get_code("2aeadc44472f2c85")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from mit-han-lab/SMEPO, and GitHub detected no licence in that repository 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. No invocation recorded.

load_hf_lm_and_tokenizer Not yet run
unverified (harvested, no recorded run)
BaohaoLiao/SAGE/eval/qwen_evaluation/model_utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from YangLing0818/SuperCorrect-llm, is pointer only (licence: NONE) · get_code("ee71a94d9dd1b8e9")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from YangLing0818/SuperCorrect-llm, and GitHub detected no licence in that repository 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. No invocation recorded.

mathd_normalize_answer Not yet run
unverified (harvested, no recorded run)
BaohaoLiao/SAGE/eval/oat_math_grader.py
code served (permissive licence) · get_code("0388e4fa2f36659e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 10 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 7 the same month, 10 later) and 0 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.

FunctionAlso present inEarliest-datedOthers named
get_weight_loader 58 other papers
15 dated earlier, 8 the same month, 33 later, 2 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2505.14362 (2025-05)
DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning
2505.20046 (2025-05)
REARANK: Reasoning Re-ranking Agent via Reinforcement Learning
and 55 more not named here
generate_config_from_args 57 other papers
8 dated earlier, 9 the same month, 37 later, 3 undated
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2509.25787 (2025-09)
Self-Evolving Vision-Language Models for Image Quality Assessment via Voting and Ranking
2510.08630 (2025-10)
ExPO-HM: Learning to Explain-then-Detect for Hateful Meme Detection
and 54 more not named here
get_dynamic_pipeline_shards 52 other papers
7 dated earlier, 8 the same month, 35 later, 2 undated
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2509.25787 (2025-09)
Self-Evolving Vision-Language Models for Image Quality Assessment via Voting and Ranking
2510.08630 (2025-10)
ExPO-HM: Learning to Explain-then-Detect for Hateful Meme Detection
and 49 more not named here
load_hf_lm_and_tokenizer 18 other papers
13 dated earlier, 5 later
2405.14365 (2024-05)
JiuZhang3.0: Efficiently Improving Mathematical Reasoning by Training Small Data Synthesis Models
2408.07471 (2024-08)
Bridging and Modeling Correlations in Pairwise Data for Direct Preference Optimization
2410.07985 (2024-10)
Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models
and 15 more not named here
generate_completions 17 other papers
12 dated earlier, 5 later
2405.14365 (2024-05)
JiuZhang3.0: Efficiently Improving Mathematical Reasoning by Training Small Data Synthesis Models
2408.07471 (2024-08)
Bridging and Modeling Correlations in Pairwise Data for Direct Preference Optimization
2410.07985 (2024-10)
Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models
and 14 more not named here
mathd_normalize_answer 12 other papers
7 dated earlier, 2 the same month, 3 later
2503.10460 (2025-03)
Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyond
2505.03335 (2025-05)
Absolute Zero: Reinforced Self-play Reasoning with Zero Data
2505.13379 (2025-05)
Thinkless: LLM Learns When to Think
and 9 more not named here
normalize_final_answer 9 other papers
5 dated earlier, 1 the same month, 3 later
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
2509.24372 (2025-09)
Evolution Strategies at Scale: LLM Fine-Tuning Beyond Reinforcement Learning
2511.13223 (2025-11)
TokenSqueeze: Performance-Preserving Compression for Reasoning LLMs
and 6 more not named here
get_weight_saver 8 other papers
1 dated earlier, 3 the same month, 3 later, 1 undated
2601.05508 (2026-01)
Enabling Stroke-Level Structural Analysis of Hieroglyphic Scripts without Language-Specific Priors
2602.03190 (2026-02)
PrAg-PO: Prompt Augmented Policy Optimization for Robust and Diverse Mathematical Reasoning
2602.05897 (2026-02)
Stop Rewarding Hallucinated Steps: Faithfulness-Aware Step-Level Reinforcement Learning for Small Reasoning Models
and 5 more not named here
repeatness 8 other papers
4 dated earlier, 1 the same month, 3 later
2509.24372 (2025-09)
Evolution Strategies at Scale: LLM Fine-Tuning Beyond Reinforcement Learning
2511.13223 (2025-11)
TokenSqueeze: Performance-Preserving Compression for Reasoning LLMs
2511.15694 (2025-11)
The Impact of Quantization on Large Reasoning Model Reinforcement Learning
and 5 more not named here
extract_boxed 1 other paper
1 later
2604.00698 (2026-04)
Learning to Hint for Reinforcement Learning

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 Policy Optimization (GRPO) has recently emerged as a practical recipe for aligning large language models with verifiable objectives. However, under sparse terminal rewards, GRPO often stalls because rollouts within a group frequently receive identical rewards, causing relative advantages to collapse and updates to vanish. We propose self-hint aligned GRPO with privileged supervision (SAGE), an on-policy reinforcement learning framework that injects privileged hints during training to reshape the rollout distribution under the same terminal verifier reward. For each prompt x, the model samples a compact hint h (e.g., a plan or decomposition) and then generates a solution τ conditioned on (x, h). Crucially, the task reward R(x, τ ) is unchanged; hints only increase within-group outcome diversity under finite sampling, preventing GRPO advantages from collapsing under sparse rewards. At test time, we set h = ∅ and deploy the no-hint policy without any privileged information. Moreover, sampling diverse selfhints serves as an adaptive curriculum that tracks the learner's bottlenecks more effectively than fixed hints from an initial policy or a stronger external model. Experiments over 6 benchmarks with 3 LLMs show that SAGE consistently outperforms GRPO, on average +2.0 on Llama-3.2-3B-Instruct, +1.2 on Qwen2.5-7B-Instruct and +1.3 on Qwen3-4B-Instruct. The code is available at https://github.com/BaohaoLiao/SAGE.

For agents

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

get_harvested_code_for_paper("2602.03143")
get_code_for_paper("2602.03143")
have("2602.03143")

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

curl https://syntology.ai/api/ran/2602.03143.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.03143.svg)](https://syntology.ai/paper/2602.03143)

Connect an agent — have() is free.