SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1912.12482 · 2019

SLM Lab: A Comprehensive Benchmark and Modular Software Framework for Reproducible Deep Reinforcement Learning

arXiv · PDF · Open in the Atlas

Code that ran

10 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; 0 honoured the contract we drafted, 1 violated it and 7 have no contract verdict; 2 failures of our instrument; 3 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/1912.12482.json · badge: /api/ran/1912.12482.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
kengz/SLM-Lab canonical (classifier label)
classifier basis: census
10 ran · 3 no recorded run
FunctionStatusWhere it lives
batch_get Ran
ran, violated the contract we drafted
kengz/SLM-Lab/slm_lab/lib/ml_util.py
code served (permissive licence) · get_code("c8a23a02f10e731a")
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 int, 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.

from collections import deque
import numpy as np
import operator

# Test case 1: Python list
arr_list = [10, 20, 30, 40, 50]
idxs_list = [0, 2, 4]

# Test case 2: Deque
arr_deque = deque([100, 200, 300, 400])
idxs_deque = [1, 3]

# Test case 3: NumPy array
arr_numpy = np.array([1.5, 2.5, 3.5, 4.5, 5.5])
idxs_numpy = np.array([0, 2, 4])

# Use the first test case for ARGS
ARGS = {
    "arr": arr_list,
    "idxs": idxs_list
}
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 int, 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
calc_srs_mean_std Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/util.py
code served (permissive licence) · get_code("1ec1b8f00401ef84")
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 2, 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 pandas as pd
import numpy as np

# Create sample pandas Series with different lengths
sr1 = pd.Series([1.0, 2.0, 3.0, 4.0, 5.0])
sr2 = pd.Series([2.0, 3.0, 4.0, 5.0, 6.0])
sr3 = pd.Series([1.5, 2.5, 3.5, 4.5, 5.5, 6.5])

sr_list = [sr1, sr2, sr3]

ARGS = {
    "sr_list": sr_list
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, 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
calc_ts_diff Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/util.py
code served (permissive licence) · get_code("1ca922f73a40acc4")
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.

from datetime import datetime

FILE_TS_FORMAT = '%Y_%m_%d_%H%M%S'

def calc_ts_diff(ts2, ts1):
    """
    Calculate the time from tss ts1 to ts2
    @param {str} ts2 Later ts in the FILE_TS_FORMAT
    @param {str} ts1 Earlier ts in the FILE_TS_FORMAT
    @returns {str} delta_t in %H:%M:%S format
    @example

    ts1 = '2017_10_17_084739'
    ts2 = '2017_10_17_084740'
    ts_diff = util.calc_ts_diff(ts2, ts1)
    # => '0:00:01'
    """
    delta_t = datetime.strptime(ts2, FILE_TS_FORMAT) - datetime.strptime(ts1, FILE_TS_FORMAT)
    return str(delta_t)


ARGS = {
    'ts2': '2017_10_17_084740',
    'ts1': '2017_10_17_084739'
}
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
center_mean Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/math_util.py
code served (permissive licence) · get_code("0a3721d2e176a664")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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

torch.manual_seed(0)

# The function center_mean takes an array-like input v
# and centers it by subtracting its mean.
# It needs to support .mean() method and subtraction.

# Create a small tensor as the input
v = torch.randn(2, 4)

ARGS = {
    "v": v
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
concat_batches Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/ml_util.py
code served (permissive licence) · get_code("7f74c5c43dd38d97")
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 numpy as np

# Create sample batches similar to what agent.memory.sample() would return
# Each batch is a dictionary with keys like 'states', 'actions', 'rewards', 'dones', etc.

# Non-episodic case: dones is a flat array
batch1 = {
    'states': np.array([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]]),
    'actions': np.array([0, 1]),
    'rewards': np.array([1.0, 2.0]),
    'dones': np.array([False, True]),
    'next_states': np.array([[9.0, 10.0, 11.0, 12.0], [13.0, 14.0, 15.0, 16.0]])
}

batch2 = {
    'states': np.array([[17.0, 18.0, 19.0, 20.0], [21.0, 22.0, 23.0, 24.0]]),
    'actions': np.array([1, 0]),
    'rewards': np.array([3.0, 4.0]),
    'dones': np.array([False, False]),
    'next_states': np.array([[25.0, 26.0, 27.0, 28.0], [29.0, 30.0, 31.0, 32.0]])
}

batches = [batch1, batch2]

ARGS = {
    'batches': batches
}
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 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
epi_done Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/ml_util.py
code served (permissive licence) · get_code("529e682f71230b22")
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.

import numpy as np

# Test case 1: scalar True (should return True)
done_scalar_true = True

# Test case 2: scalar False (should return False)
done_scalar_false = False

# Test case 3: numpy array (not scalar, should return False)
done_array = np.array([True, False])

# The function checks if done is scalar AND done is True
# We'll test with a scalar boolean value
ARGS = {
    "done": done_scalar_true
}
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 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
format_metrics Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/util.py
code served (permissive licence) · get_code("2b7e1db927d58735")
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)

# Construct a comprehensive metrics dictionary that exercises all code paths
metrics = {
    'frame': 1.5e7,
    'total_reward': 42.567,
    'total_reward_ma': 38.234,
    'loss': 0.00123,
    'accuracy': 0.9876,
    'episode': 100,
    'counter': 5,
    'nan_value': float('nan'),
}

# Add a torch tensor that has .item() method
tensor_metric = torch.tensor(3.14159)
metrics['torch_value'] = tensor_metric

# Add a numpy scalar
metrics['numpy_value'] = np.float32(2.71828)

# Add a regular integer
metrics['steps'] = 1000

# Add a value that is an integer float (should use :g formatting)
metrics['integer_float'] = 5.0

ARGS = {
    'metrics': metrics
}
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 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
set_from_cli Ran
ran on a synthesized input
kengz/SLM-Lab/slm_lab/lib/env_var.py
code served (permissive licence) · get_code("b68ae619f0cef90a")
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 os

# Set up environment for testing
os.environ.clear()

# Define arguments for set_from_cli function
ARGS = {
    'render': True,
    'log_level': 'DEBUG',
    'optimize_perf': False,
    'cuda_offset': 2,
    'profile': True,
    'log_extra': True,
    'upload_hf': False,
    'mode': 'train'
}
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
symexp Ran
ran; our contract draft was wrong, not the code
kengz/SLM-Lab/slm_lab/lib/math_util.py
code served (permissive licence) · get_code("840248f6e820c2cf")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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 symexp takes a single parameter x which should be a torch tensor
# It applies element-wise operations: torch.sign, torch.abs, torch.exp
# So x needs to be a tensor

x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
symlog Ran
ran; our contract draft was wrong, not the code
kengz/SLM-Lab/slm_lab/lib/math_util.py
code served (permissive licence) · get_code("bb21890c6aa0c40e")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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)

ARGS = {
    "x": torch.randn(2, 4),
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
critical Not yet run
unverified (harvested, no recorded run)
kengz/SLM-Lab/slm_lab/lib/logger.py
code served (permissive licence) · get_code("dc69a6b4b1eb2489")
How it ran

no invocation recorded.

debug Not yet run
unverified (harvested, no recorded run)
kengz/SLM-Lab/slm_lab/lib/logger.py
code served (permissive licence) · get_code("b5c2a41c3a816ab2")
How it ran

no invocation recorded.

error Not yet run
unverified (harvested, no recorded run)
kengz/SLM-Lab/slm_lab/lib/logger.py
code served (permissive licence) · get_code("ac80a904d8e853b1")
How it ran

no invocation 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, 4 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 4 later) and 9 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
batch_get 1 other paper
1 later
2108.13679 (2021-08)
Task-Oriented Dialogue System as Natural Language Generation
calc_ts_diff 1 other paper
1 later
2108.13679 (2021-08)
Task-Oriented Dialogue System as Natural Language Generation
symexp 1 other paper
1 later
2310.17805 (2023-10)
Reward Scale Robustness for Proximal Policy Optimization via DreamerV3 Tricks
symlog 1 other paper
1 later
2603.18202 (2026-03)
R2-Dreamer: Redundancy-Reduced World Models without Decoders or Augmentation

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

We introduce SLM Lab, a software framework for reproducible reinforcement learning (RL) research. SLM Lab implements a number of popular RL algorithms, provides synchronous and asynchronous parallel experiment execution, hyperparameter search, and result analysis. RL algorithms in SLM Lab are implemented in a modular way such that differences in algorithm performance can be confidently ascribed to differences between algorithms, not between implementations. In this work we present the design choices behind SLM Lab and use it to produce a comprehensive single-codebase RL algorithm benchmark. In addition, as a consequence of SLM Lab's modular design, we introduce and evaluate a discrete-action variant of the Soft Actor-Critic algorithm (Haarnoja et al., 2018) and a hybrid synchronous/asynchronous training method for RL agents.

For agents

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

get_harvested_code_for_paper("1912.12482")
get_code_for_paper("1912.12482")
have("1912.12482")

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

curl https://syntology.ai/api/ran/1912.12482.json

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

[![Syntology run record](https://syntology.ai/api/ran/1912.12482.svg)](https://syntology.ai/paper/1912.12482)

Connect an agent — have() is free.