SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.06547 · NeurIPS · 2020

Avoiding Side Effects in Complex Environments

Alexander Turner, Neale Ratzlaff, Prasad Tadepalli

arXiv · PDF · Open in the Atlas

Code that ran

5 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 5 have no contract verdict; 0 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/2006.06547.json · badge: /api/ran/2006.06547.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
aseembits93/avoiding-side-effects community (classifier label)
classifier basis not recorded on the link
5 ran · 3 no recorded run
FunctionStatusWhere it lives
coinflip Ran
ran on a synthesized input
aseembits93/avoiding-side-effects/safelife/random.py
code served (permissive licence) · get_code("0737b2dfc6f56566")
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 (5,) 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

# Set random seed for reproducibility
np.random.seed(42)

# Define ARGS dict with all parameters
ARGS = {
    "p": 0.7,
    "n": None
}

# Test case 1: Single coin flip
ARGS_single = {
    "p": 0.5,
    "n": None
}

# Test case 2: Multiple coin flips (1D array)
ARGS_array_1d = {
    "p": 0.3,
    "n": 10
}

# Test case 3: Multiple coin flips (multi-dimensional array)
ARGS_array_2d = {
    "p": 0.6,
    "n": (2, 4)
}

# Use the basic case as the main ARGS
ARGS = {
    "p": 0.75,
    "n": 5
}
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 (5,) — shape (5,) 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
load_benchmarks Ran
ran on a synthesized input
aseembits93/avoiding-side-effects/safelife/benchmarking.py
code served (permissive licence) · get_code("9b6b794b330f3d9f")
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 yaml
import numpy as np
import tempfile
import os

# Create a temporary YAML file with benchmark data
temp_dir = tempfile.mkdtemp()
logfile_path = os.path.join(temp_dir, "benchmarks.yaml")

# Create sample benchmark data
benchmark_data = [
    {
        'episode': 1,
        'reward': 100.5,
        'steps': 50,
        'side effects': {
            'energy_used': [10.2, 5.1],
            'time_elapsed': [2.3, 1.1]
        }
    },
    {
        'episode': 2,
        'reward': 105.3,
        'steps': 48,
        'side effects': {
            'energy_used': [9.8, 4.9],
            'time_elapsed': [2.1, 1.0]
        }
    },
    {
        'episode': 3,
        'reward': 98.7,
        'steps': 52,
        'side effects': {
            'energy_used': [10.5, 5.3],
            'time_elapsed': [2.4, 1.2]
        }
    }
]

# Write the YAML file
with open(logfile_path, 'w') as f:
    yaml.dump(benchmark_data, f)

ARGS = {
    'logfile': logfile_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, yaml — 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
safelife_cnn Ran
ran on a synthesized input
aseembits93/avoiding-side-effects/training/models.py
code served (permissive licence) · get_code("bb2ddcd53c04fdc1")
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. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "input_shape": (25, 25, 3)
}
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 tuple — a tuple output has no array shape of its own; none is 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
safelife_cnn Ran
ran on a synthesized input
aseembits93/avoiding-side-effects/training/models_safelife.py
code served (permissive licence) · get_code("9f7a5c4fbc840fc9")
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. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "input_shape": (25, 25, 3),
}
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 tuple — a tuple output has no array shape of its own; none is 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
signed_sqrt Ran
ran on a synthesized input
aseembits93/avoiding-side-effects/training/models.py
code served (permissive licence) · get_code("0047a0112d7955f1")
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 signed_sqrt takes a single parameter 'x' which should be a tensor
# It computes the sign, takes absolute value, applies sqrt, and multiplies back
# So x needs to be a torch.Tensor with both positive and negative values

x = torch.randn(2, 4)  # Small tensor with both positive and negative values

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
build_fence Not yet run
unverified (harvested, no recorded run)
aseembits93/avoiding-side-effects/safelife/proc_gen.py
code served (permissive licence) · get_code("b618888381b55599")
How it ran

no invocation recorded.

make_partioned_regions Not yet run
unverified (harvested, no recorded run)
aseembits93/avoiding-side-effects/safelife/proc_gen.py
code served (permissive licence) · get_code("1cef04eaa5c76483")
How it ran

no invocation recorded.

recenter_view Not yet run
unverified (harvested, no recorded run)
aseembits93/avoiding-side-effects/safelife/helper_utils.py
code served (permissive licence) · get_code("3ad72d4e41ca99d6")
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, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 0 later) and 4 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
build_fence 1 other paper
1 dated earlier
1912.01217 (2019-12)
SafeLife 1.0: Exploring Side Effects in Complex Environments
coinflip 1 other paper
1 dated earlier
1912.01217 (2019-12)
SafeLife 1.0: Exploring Side Effects in Complex Environments
make_partioned_regions 1 other paper
1 dated earlier
1912.01217 (2019-12)
SafeLife 1.0: Exploring Side Effects in Complex Environments
recenter_view 1 other paper
1 dated earlier
1912.01217 (2019-12)
SafeLife 1.0: Exploring Side Effects in Complex Environments

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

Reward function specification can be difficult. Rewarding the agent for making a widget may be easy, but penalizing the multitude of possible negative side effects is hard. In toy environments, Attainable Utility Preservation (AUP) avoided side effects by penalizing shifts in the ability to achieve randomly generated goals [22]. We scale this approach to large, randomly generated environments based on Conway's Game of Life. By preserving optimal value for a single randomly generated reward function, AUP incurs modest overhead while leading the agent to complete the specified task and avoid many side effects. Videos and code are available at https://avoiding-side-effects.github.io/.

For agents

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

get_harvested_code_for_paper("2006.06547")
get_code_for_paper("2006.06547")
have("2006.06547")

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

curl https://syntology.ai/api/ran/2006.06547.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.06547.svg)](https://syntology.ai/paper/2006.06547)

Connect an agent — have() is free.