SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.04597 · 2026

Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning

Lei Huang, Junjie Yang, Xiang Cheng, Guobin Shen, Bing Qin, Yuxuan Gu, Xiaocheng Feng, Xing Yu, Chenxiao Zhao

arXiv · PDF · Open in the Atlas

Code that ran

7 of 8 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; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 7 failures of our instrument; 1 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/2603.04597.json · badge: /api/ran/2603.04597.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
LuckyyySTA/GOLF canonical (classifier label)
classifier basis: census
7 ran · 1 no recorded run
FunctionStatusWhere it lives
accuracy_reward Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
LuckyyySTA/GOLF/golf/verl/verl/adaptive_mix_src/reward_with_format.py
code served (permissive licence) · get_code("b7250010ea994bcd")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 torch
import numpy as np

torch.manual_seed(0)

# Mock the math_verify module components since they're imported in the function
class MockLatexExtractionConfig:
    def __init__(self, normalization_config=None, boxed_match_priority=None, try_extract_without_anchor=None):
        self.normalization_config = normalization_config
        self.boxed_match_priority = boxed_match_priority
        self.try_extract_without_anchor = try_extract_without_anchor

class MockNormalizationConfig:
    def __init__(self, nits=False, malformed_operators=False, basic_latex=True, equations=True, boxed='all', units=True):
        self.nits = nits
        self.malformed_operators = malformed_operators
        self.basic_latex = basic_latex
        self.equations = equations
        self.boxed = boxed
        self.units = units

def mock_parse(content, extraction_mode='first_match', extraction_config=None):
    """Mock parse function that returns a non-empty list for valid content"""
    if content and len(content) > 0:
        return [{"parsed": content}]
    return []

def mock_verify(answer_parsed, gold_parsed):
    """Mock verify function that returns a boolean"""
    if answer_parsed and gold_parsed:
        return True
    return False

# Inject mocks into a fake math_verify module
import sys
from types import ModuleType

math_verify = ModuleType('math_verify')
math_verify.LatexExtractionConfig = MockLatexExtractionConfig
math_verify.parse = mock_parse
math_verify.verify = mock_verify
sys.modules['math_verify'] = math_verify

latex2sympy2_extended = ModuleType('latex2sympy2_extended')
latex2sympy2_extended.NormalizationConfig = MockNormalizationConfig
sys.modules['latex2sympy2_extended'] = latex2sympy2_extended

# Now define the arguments
ARGS = {
    'completions': [
        r"The answer is $x = 5$",
        r"The result is $y = 10$"
    ],
    'solution': [
        r"$x = 5$",
        r"$y = 10$"
    ]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
latex2sympy2_extended, math_verify, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
compute_ce_dpo_loss_rm Ran
ran; our contract draft was wrong, not the code
LuckyyySTA/GOLF/golf/verl/recipe/prime/prime_core_algos.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("f220595adfc5e83c")
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.

compute_detach_dpo_loss_rm Ran
ran; our contract draft was wrong, not the code
LuckyyySTA/GOLF/golf/verl/recipe/prime/prime_core_algos.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("4d91e4383716e42c")
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.

compute_online_dpo_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
LuckyyySTA/GOLF/golf/verl/recipe/spin/core_algos.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("f19825983a523003")
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.

compute_onlinedpo_pref Ran
ran; our contract draft was wrong, not the code
LuckyyySTA/GOLF/golf/verl/recipe/spin/core_algos.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("41b6e81dc3ecba79")
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_kl_controller Ran
ran; our contract draft was wrong, not the code
LuckyyySTA/GOLF/golf/verl/recipe/spin/core_algos.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from LeapLabTHU/OPDVR, is pointer only (licence: NONE) · get_code("757d109d90121142")
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.

reward_func Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
LuckyyySTA/GOLF/golf/verl/recipe/r1/reward_score.py
code served (permissive licence) · get_code("c30b2a96103abd2c")
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 MagicMock. 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 torch
import numpy as np

torch.manual_seed(0)

# Mock the recipe.r1.tasks modules to avoid import errors
import sys
from unittest.mock import MagicMock

mock_math = MagicMock()
mock_math.compute_score = lambda sol, gt: 0.85

mock_gpqa = MagicMock()
mock_gpqa.compute_score = lambda sol, gt: 0.92

mock_livecodebench = MagicMock()
mock_livecodebench.compute_score = lambda sol, gt: 0.78

sys.modules['recipe'] = MagicMock()
sys.modules['recipe.r1'] = MagicMock()
sys.modules['recipe.r1.tasks'] = MagicMock()
sys.modules['recipe.r1.tasks.math'] = mock_math
sys.modules['recipe.r1.tasks.gpqa'] = mock_gpqa
sys.modules['recipe.r1.tasks.livecodebench'] = mock_livecodebench

# Define arguments for reward_func
ARGS = {
    'data_source': 'Maxwell-Jia/AIME_2024',
    'solution_str': 'The answer is 42.',
    'ground_truth': '42',
    'extra_info': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, recipe, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MagicMock — a MagicMock output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MagicMock output the recorded ndim 1 and dtype family None 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
math_verify_reward_fn Not yet run
unverified (harvested, no recorded run)
LuckyyySTA/GOLF/golf/verl/verl/adaptive_mix_src/reward_with_format.py
code served (permissive licence) · get_code("e90d214897d14000")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 6 the same month, 6 later) and 2 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
compute_ce_dpo_loss_rm 46 other papers
19 dated earlier, 6 the same month, 18 later, 3 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
and 43 more not named here
compute_detach_dpo_loss_rm 44 other papers
17 dated earlier, 7 the same month, 17 later, 3 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
and 41 more not named here
compute_online_dpo_loss 38 other papers
13 dated earlier, 5 the same month, 17 later, 3 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
and 35 more not named here
get_kl_controller 38 other papers
13 dated earlier, 5 the same month, 17 later, 3 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
and 35 more not named here
compute_onlinedpo_pref 30 other papers
7 dated earlier, 5 the same month, 15 later, 3 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2410.02992 (2024-10)
Learning to Better Search with Language Models via Guided Reinforced Self-Training
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
and 27 more not named here
reward_func 28 other papers
13 dated earlier, 3 the same month, 10 later, 2 undated
2409.19256 (2024-09)
HybridFlow: A Flexible and Efficient RLHF Framework
2504.16084 (2025-04)
TTRL: Test-Time Reinforcement Learning
2505.14362 (2025-05)
DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning
and 25 more not named here

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

Large language models (LLMs) typically receive diverse natural language (NL) feedback through interaction with the environment. However, current reinforcement learning (RL) algorithms rely solely on scalar rewards, leaving the rich information in NL feedback underutilized and leading to inefficient exploration. In this work, we propose GOLF, an RL framework that explicitly exploits GrOup-level Language Feedback to guide targeted exploration through actionable refinements. GOLF aggregates two complementary feedback sources: (i) external critiques that pinpoint errors or propose targeted fixes, and (ii) intra-group attempts that supply alternative partial ideas and diverse failure patterns. These group-level feedbacks are aggregated to produce high-quality refinements, which are adaptively injected into training as off-policy scaffolds to provide targeted guidance in sparse-reward regions. Meanwhile, GOLF jointly optimizes generation and refinement within a unified RL loop, creating a virtuous cycle that continuously improves both capabilities. Experiments on both verifiable and non-verifiable benchmarks show that GOLF achieves superior performance and exploration efficiency, achieving 2.2× improvements in sample efficiency compared to RL methods trained solely on scalar rewards. Code is available at https: //github.com/LuckyyySTA/GOLF

For agents

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

get_harvested_code_for_paper("2603.04597")
get_code_for_paper("2603.04597")
have("2603.04597")

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

curl https://syntology.ai/api/ran/2603.04597.json

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

[![Syntology run record](https://syntology.ai/api/ran/2603.04597.svg)](https://syntology.ai/paper/2603.04597)

Connect an agent — have() is free.