SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2401.14758 · ICLR · 2024

Off-Policy Primal-Dual Safe Reinforcement Learning

Dong Wang, Zifan Wu, Bo Tang, Qian Lin, Yu Chao, Shangqin Mao, Qianlong Xie, Xingxing Wang

arXiv · PDF · Open in the Atlas

Code that ran

13 of 22 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 4 honoured the contract we drafted, 2 violated it and 1 has no contract verdict; 6 failures of our instrument; 9 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2401.14758.json · badge: /api/ran/2401.14758.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
ZifanWu/CAL canonical (classifier label)
classifier basis: census
6 ran · 1 no recorded run
AlgTUDelft/WCSAC canonical (classifier label)
classifier basis: census
5 ran · 5 no recorded run
zmsn-2077/CUP-safe-rl community
Papers with Code archive
2 ran · 3 no recorded run
FunctionStatusWhere it lives
CUP Ran
ran on a synthesized input · constructed an object; no result was computed
zmsn-2077/CUP-safe-rl/algorithm/cup.py
pointer only (licence: NONE) · get_code("e47f49c341d79904")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zmsn-2077/CUP-safe-rl 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.

colorize Ran
ran; our contract draft was wrong, not the code
AlgTUDelft/WCSAC/wc_sac/utils/logx.py
pointer only (licence: MIT); the stored copy get_code serves, from milanganai/milanganai.github.io, is pointer only (licence: NONE) · get_code("7e36f1c357c4618d")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from milanganai/milanganai.github.io, 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.

convert_json Ran
ran; our contract draft was wrong, not the code
AlgTUDelft/WCSAC/wc_sac/utils/serialization_utils.py
code served (permissive licence) · get_code("fbc505a5dac82646")
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) and returned dict. 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 json

torch.manual_seed(0)

# Test case 1: JSON-serializable object (int)
obj1 = 42

# Test case 2: dict with nested structures
obj2 = {
    "name": "test",
    "values": [1, 2, 3],
    "nested": {"key": "value"}
}

# Test case 3: tuple
obj3 = (1, 2, "three")

# Test case 4: list with mixed types
obj4 = [1, "two", 3.14, True, None]

# Test case 5: object with __name__ attribute (function)
def my_function():
    pass
obj5 = my_function

# Test case 6: object with __dict__ (custom class instance)
class CustomClass:
    def __init__(self):
        self.attr1 = "value1"
        self.attr2 = 42

obj6 = CustomClass()

# Test case 7: tensor (not JSON serializable)
obj7 = torch.tensor([1.0, 2.0, 3.0, 4.0])

# Test case 8: complex nested structure
obj8 = {
    "tensors": [torch.randn(2, 4)],
    "objects": [CustomClass()],
    "functions": [my_function],
    "primitives": [1, "two", 3.14]
}

# Test case 9: lambda function (should convert to string)
obj9 = lambda x: x + 1

# Test case 10: empty dict
obj10 = {}

ARGS = {
    "obj": obj8
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
create_log_gaussian Ran
ran; our contract draft was wrong, not the code
ZifanWu/CAL/agent/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from roythuly/ompo, is pointer only (licence: NONE) · get_code("2526cfd8d5bf5d06")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from roythuly/ompo, 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.

flat_concat Ran
ran, honoured the contract we drafted
AlgTUDelft/WCSAC/wc_sac/utils/mpi_tf.py
pointer only (licence: MIT); the stored copy get_code serves, from milanganai/milanganai.github.io, is pointer only (licence: NONE) · get_code("f29202d8108bccd3")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from milanganai/milanganai.github.io, 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.

gaussian_kl Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zmsn-2077/CUP-safe-rl/algorithm/cup.py
pointer only (licence: NONE) · get_code("5b17e81927faf55e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zmsn-2077/CUP-safe-rl 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_threshold Ran
ran, honoured the contract we drafted
ZifanWu/CAL/env/constraints.py
code served (permissive licence) · get_code("cb396863dfa04b08")
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 float, 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.

import torch
import numpy as np

# The function get_threshold takes an env parameter and an optional constraint parameter
# It looks up the env in a dictionary based on the constraint type
# For 'velocity' constraint, valid envs are: 'Ant-v3', 'HalfCheetah-v3', 'Hopper-v3', 'Humanoid-v3'
# For 'safetygym' constraint, valid envs are: 'Safexp-CarButton1-v0', etc.

ARGS = {
    'env': 'Ant-v3',
    'constraint': 'velocity'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
is_json_serializable Ran
ran, violated the contract we drafted
AlgTUDelft/WCSAC/wc_sac/utils/serialization_utils.py
code served (permissive licence) · get_code("9ca8b8e45f7fbebf")
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) 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.

import json

def is_json_serializable(v):
    try:
        json.dumps(v)
        return True
    except:
        return False

# Test with various values
ARGS = {
    "v": {"key": "value", "number": 42, "list": [1, 2, 3]}
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
logsumexp Ran
ran; our contract draft was wrong, not the code
ZifanWu/CAL/agent/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from roythuly/ompo, is pointer only (licence: NONE) · get_code("09556d56c8b73050")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from roythuly/ompo, 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 Ran
ran, honoured the contract we drafted
ZifanWu/CAL/env/safety-gym/safety_gym/bench/bench_utils.py
code served (permissive licence) · get_code("d7740644bf4b7de1")
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 tuple, 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 numpy as np
import json
import tempfile
import os

# Create a temporary directory and a mock characteristic_scores.json file
temp_dir = tempfile.mkdtemp()
safety_gym_bench_dir = os.path.join(temp_dir, 'safety_gym', 'bench')
os.makedirs(safety_gym_bench_dir, exist_ok=True)

# Create mock characteristic scores
characteristic_scores = {
    "pointgoal1": {
        "Ret": 100.0,
        "Cost": 50.0,
        "CostRate": 0.1
    }
}

scores_file = os.path.join(safety_gym_bench_dir, 'characteristic_scores.json')
with open(scores_file, 'w') as f:
    json.dump(characteristic_scores, f)

# Change to temp directory so the function can find the file
original_dir = os.getcwd()
os.chdir(temp_dir)

ARGS = {
    'env': 'Safexp-PointGoal1-v0',
    'ret': 75.5,
    'cost': 30.0,
    'costrate': 0.08,
    'cost_limit': 25,
    'round': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple 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
quat2zalign Ran
ran, violated the contract we drafted
ZifanWu/CAL/env/safety-gym/safety_gym/envs/engine.py
code served (permissive licence) · get_code("4ac8031d738ec401")
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 float, 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.

import torch

torch.manual_seed(0)

# The function quat2zalign takes a quaternion (4 values: a, b, c, d)
# and performs arithmetic operations on them
# We need to provide a sequence/tuple of 4 numeric values

quat = (0.5, 0.3, 0.2, 0.8)

ARGS = {
    "quat": quat
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
setup_logger_kwargs Ran
ran; our contract draft was wrong, not the code
AlgTUDelft/WCSAC/wc_sac/utils/run_utils.py
code served (permissive licence) · get_code("cc9f89f4007acda0")
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 dict. 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 time
import os.path as osp
import tempfile

# Create a temporary directory to use as data_dir
temp_dir = tempfile.mkdtemp()

ARGS = {
    'exp_name': 'test_experiment',
    'seed': 42,
    'data_dir': temp_dir,
    'datestamp': True
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
theta2vec Ran
ran, honoured the contract we drafted
ZifanWu/CAL/env/safety-gym/safety_gym/envs/engine.py
code served (permissive licence) · get_code("39708d407531ef32")
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 ndarray, ndim 1, dtype family float, shape (3,) 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.

import numpy as np

# theta2vec takes a single parameter: theta (an angle in radians)
# The function converts it to a unit vector using np.cos and np.sin

# Create a valid angle in radians
theta_value = np.pi / 4  # 45 degrees, a simple test value

ARGS = {
    "theta": theta_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (3,) — shape (3,) 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
assign_params_from_flat Not yet run
unverified (harvested, no recorded run)
AlgTUDelft/WCSAC/wc_sac/utils/mpi_tf.py
pointer only (licence: MIT); the stored copy get_code serves, from milanganai/milanganai.github.io, is pointer only (licence: NONE) · get_code("042875cf711accd0")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from milanganai/milanganai.github.io, 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.

get_vars Not yet run
unverified (harvested, no recorded run)
AlgTUDelft/WCSAC/wc_sac/sac/wcsac.py
code served (permissive licence) · get_code("a2580d7c4a763348")
How it ran

no invocation recorded.

graph_detach Not yet run
unverified (harvested, no recorded run)
zmsn-2077/CUP-safe-rl/algorithm/cup.py
pointer only (licence: NONE) · get_code("7d8c58ad49e14ee5")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zmsn-2077/CUP-safe-rl 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.

init_weights Not yet run
unverified (harvested, no recorded run)
ZifanWu/CAL/agent/model.py
code served (permissive licence) · get_code("0661bf42848a4988")
How it ran

no invocation recorded.

mlp Not yet run
unverified (harvested, no recorded run)
AlgTUDelft/WCSAC/wc_sac/sac/wcsac.py
code served (permissive licence) · get_code("e6f12e62f88b74b9")
How it ran

no invocation recorded.

placeholder Not yet run
unverified (harvested, no recorded run)
AlgTUDelft/WCSAC/wc_sac/sac/wcsac.py
code served (permissive licence) · get_code("17b03c462ce92576")
How it ran

no invocation recorded.

println Not yet run
unverified (harvested, no recorded run)
zmsn-2077/CUP-safe-rl/algorithm/cup.py
pointer only (licence: NONE) · get_code("a3c2d567fb278fea")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zmsn-2077/CUP-safe-rl 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.

restore_tf_graph Not yet run
unverified (harvested, no recorded run)
AlgTUDelft/WCSAC/wc_sac/utils/logx.py
code served (permissive licence) · get_code("006604b1b94a9fe7")
How it ran

no invocation recorded.

to_dytype_device Not yet run
unverified (harvested, no recorded run)
zmsn-2077/CUP-safe-rl/algorithm/cup.py
pointer only (licence: NONE) · get_code("495130bd57c2f5c1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, zmsn-2077/CUP-safe-rl 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.

Also present in other papers' code

Of 22 samples lifted from this paper's linked repositories, 22 carry a function identity and 0 do not. Of the 22 distinct functions with an identity, 14 are also present in other papers whose code we hold (14 with at least one paper dated earlier, 2 the same month, 7 later) and 8 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
colorize 17 other papers
11 dated earlier, 6 later
1802.09477 (2018-02)
Addressing Function Approximation Error in Actor-Critic Methods
1909.12255 (2019-09)
Harnessing Structures for Value-Based Planning and Reinforcement Learning
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
and 14 more not named here
is_json_serializable 9 other papers
6 dated earlier, 3 later
1802.09477 (2018-02)
Addressing Function Approximation Error in Actor-Critic Methods
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
2106.09110 (2021-06)
Safe Reinforcement Learning Using Advantage-Based Intervention
and 6 more not named here
logsumexp 9 other papers
7 dated earlier, 1 the same month, 1 later
1906.02771 (2019-06)
Improving Exploration in Soft-Actor-Critic with Normalizing Flows Policies
1911.02256 (2019-11)
A Divergence Minimization Perspective on Imitation Learning Methods
2004.14451 (2020-04)
Pragmatic Issue-Sensitive Image Captioning
and 6 more not named here
convert_json 7 other papers
5 dated earlier, 2 later
1802.09477 (2018-02)
Addressing Function Approximation Error in Actor-Critic Methods
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
2106.09110 (2021-06)
Safe Reinforcement Learning Using Advantage-Based Intervention
and 4 more not named here
create_log_gaussian 7 other papers
5 dated earlier, 1 the same month, 1 later
1906.02771 (2019-06)
Improving Exploration in Soft-Actor-Critic with Normalizing Flows Policies
2010.08830 (2020-10)
MESA: Boost Ensemble Imbalanced Learning with MEta-SAmpler
2010.09163 (2020-10)
D2RL: Deep Dense Architectures in Reinforcement Learning
and 4 more not named here
flat_concat 6 other papers
4 dated earlier, 2 later
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
2210.07573 (2022-10)
Model-based Safe Deep Reinforcement Learning via a Constrained Proximal Policy Optimization Algorithm
2302.00808 (2023-02)
ACPO: A Policy Optimization Algorithm for Average MDPs with Constraints
and 3 more not named here
assign_params_from_flat 4 other papers
3 dated earlier, 1 later
2006.07442 (2020-06)
Self-Imitation Learning via Generalized Lower Bound Q-learning
2210.07573 (2022-10)
Model-based Safe Deep Reinforcement Learning via a Constrained Proximal Policy Optimization Algorithm
2302.00808 (2023-02)
ACPO: A Policy Optimization Algorithm for Average MDPs with Constraints
and 1 more not named here
get_vars 4 other papers
4 dated earlier
1802.09477 (2018-02)
Addressing Function Approximation Error in Actor-Critic Methods
1812.02690 (2018-12)
Provably Efficient Maximum Entropy Exploration
1910.07224 (2019-10)
Teacher algorithms for curriculum learning of Deep RL in continuously parameterized environments
and 1 more not named here
normalize 4 other papers
4 dated earlier
2201.12427 (2022-01)
Towards Safe Reinforcement Learning with a Safety Editor Policy
2303.03339 (2023-03)
Reducing Safety Interventions in Provably Safe Reinforcement Learning
2306.00035 (2023-06)
ROSARL: Reward-Only Safe Reinforcement Learning
and 1 more not named here
placeholder 4 other papers
4 dated earlier
1802.09477 (2018-02)
Addressing Function Approximation Error in Actor-Critic Methods
1812.02690 (2018-12)
Provably Efficient Maximum Entropy Exploration
1910.07224 (2019-10)
Teacher algorithms for curriculum learning of Deep RL in continuously parameterized environments
and 1 more not named here
quat2zalign 3 other papers
3 dated earlier
2201.12427 (2022-01)
Towards Safe Reinforcement Learning with a Safety Editor Policy
2303.03339 (2023-03)
Reducing Safety Interventions in Provably Safe Reinforcement Learning
2311.02227 (2023-11)
State-Wise Safe Reinforcement Learning With Pixel Observations
theta2vec 3 other papers
3 dated earlier
2201.12427 (2022-01)
Towards Safe Reinforcement Learning with a Safety Editor Policy
2303.03339 (2023-03)
Reducing Safety Interventions in Provably Safe Reinforcement Learning
2311.02227 (2023-11)
State-Wise Safe Reinforcement Learning With Pixel Observations
init_weights 2 other papers
2 dated earlier
2201.11927 (2022-01)
Constrained Variational Policy Optimization for Safe Reinforcement Learning
2210.07573 (2022-10)
Model-based Safe Deep Reinforcement Learning via a Constrained Proximal Policy Optimization Algorithm
restore_tf_graph 1 other paper
1 dated earlier
2302.00808 (2023-02)
ACPO: A Policy Optimization Algorithm for Average MDPs with Constraints

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

Primal-dual safe RL methods commonly perform iterations between the primal update of the policy and the dual update of the Lagrange Multiplier. Such a training paradigm is highly susceptible to the error in cumulative cost estimation since this estimation serves as the key bond connecting the primal and dual update processes. We show that this problem causes significant underestimation of cost when using off-policy methods, leading to the failure to satisfy the safety constraint. To address this issue, we propose conservative policy optimization, which learns a policy in a constraint-satisfying area by considering the uncertainty in cost estimation. This improves constraint satisfaction but also potentially hinders reward maximization. We then introduce local policy convexification to help eliminate such suboptimality by gradually reducing the estimation uncertainty. We provide theoretical interpretations of the joint coupling effect of these two ingredients and further verify them by extensive experiments. Results on benchmark tasks show that our method not only achieves an asymptotic performance comparable to state-of-the-art on-policy methods while using much fewer samples, but also significantly reduces constraint violation during training. Our code is available at https://github.com/ZifanWu/CAL.

For agents

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

get_harvested_code_for_paper("2401.14758")
get_code_for_paper("2401.14758")
have("2401.14758")

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

curl https://syntology.ai/api/ran/2401.14758.json

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

[![Syntology run record](https://syntology.ai/api/ran/2401.14758.svg)](https://syntology.ai/paper/2401.14758)

Connect an agent — have() is free.