SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.16801 · 2024

Craftax: A Lightning-Fast Benchmark for Open-Ended Reinforcement Learning

arXiv · PDF · Open in the Atlas

Code that ran

2 of 16 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, 0 violated it and 0 have no contract verdict; 1 failure of our instrument; 14 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/2402.16801.json · badge: /api/ran/2402.16801.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
michaeltmatthews/craftax canonical (classifier label)
classifier basis: census
2 ran · 14 no recorded run
FunctionStatusWhere it lives
achievement_mapping Ran
ran, honoured the contract we drafted
michaeltmatthews/craftax/craftax/craftax/constants.py
code served (permissive licence) · get_code("25c2020d2f7ef242")
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 int, ndim 0, dtype family int, 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
from enum import Enum

class Achievement(Enum):
    COLLECT_WOOD = 0
    PLACE_TABLE = 1
    EAT_COW = 2
    COLLECT_SAPLING = 3
    COLLECT_DRINK = 4
    MAKE_WOOD_PICKAXE = 5
    MAKE_WOOD_SWORD = 6
    PLACE_PLANT = 7
    DEFEAT_ZOMBIE = 8
    COLLECT_STONE = 9
    PLACE_STONE = 10
    EAT_PLANT = 11
    DEFEAT_SKELETON = 12
    MAKE_STONE_PICKAXE = 13
    MAKE_STONE_SWORD = 14
    WAKE_UP = 15
    PLACE_FURNACE = 16
    COLLECT_COAL = 17
    COLLECT_IRON = 18
    COLLECT_DIAMOND = 19
    MAKE_IRON_PICKAXE = 20
    MAKE_IRON_SWORD = 21
    MAKE_ARROW = 22
    MAKE_TORCH = 23
    PLACE_TORCH = 24
    COLLECT_SAPPHIRE = 54
    COLLECT_RUBY = 59
    MAKE_DIAMOND_PICKAXE = 60
    MAKE_DIAMOND_SWORD = 25
    MAKE_IRON_ARMOUR = 26
    MAKE_DIAMOND_ARMOUR = 27
    ENTER_GNOMISH_MINES = 28
    ENTER_DUNGEON = 29
    ENTER_SEWERS = 30
    ENTER_VAULT = 31
    ENTER_TROLL_MINES = 32
    ENTER_FIRE_REALM = 33
    ENTER_ICE_REALM = 34
    ENTER_GRAVEYARD = 35
    DEFEAT_GNOME_WARRIOR = 36
    DEFEAT_GNOME_ARCHER = 37
    DEFEAT_ORC_SOLIDER = 38
    DEFEAT_ORC_MAGE = 39
    DEFEAT_LIZARD = 40
    DEFEAT_KOBOLD = 41
    DEFEAT_KNIGHT = 65
    DEFEAT_ARCHER = 66
    DEFEAT_TROLL = 42
    DEFEAT_DEEP_THING = 43
    DEFEAT_PIGMAN = 44
    DEFEAT_FIRE_ELEMENTAL = 45
    DEFEAT_FROST_TROLL = 46
    DEFEAT_ICE_ELEMENTAL = 47
    DAMAGE_NECROMANCER = 48
    DEFEAT_NECROMANCER = 49
    EAT_BAT = 50
    EAT_SNAIL = 51
    FIND_BOW = 52
    FIRE_BOW = 53
    LEARN_FIREBALL = 55
    CAST_FIREBALL = 56
    LEARN_ICEBALL = 57
    CAST_ICEBALL = 58
    OPEN_CHEST = 61
    DRINK_POTION = 62
    ENCHANT_SWORD = 63
    ENCHANT_ARMOUR = 64

INTERMEDIATE_ACHIEVEMENTS = [Achievement.COLLECT_SAPPHIRE.value, Achievement.COLLECT_RUBY.value, Achievement.MAKE_DIAMOND_PICKAXE.value, Achievement.MAKE_DIAMOND_SWORD.value, Achievement.MAKE_IRON_ARMOUR.value, Achievement.MAKE_DIAMOND_ARMOUR.value, Achievement.ENTER_GNOMISH_MINES.value, Achievement.ENTER_DUNGEON.value, Achievement.DEFEAT_GNOME_WARRIOR.value, Achievement.DEFEAT_GNOME_ARCHER.value, Achievement.DEFEAT_ORC_SOLIDER.value, Achievement.DEFEAT_ORC_MAGE.value, Achievement.EAT_BAT.value, Achievement.EAT_SNAIL.value, Achievement.FIND_BOW.value, Achievement.FIRE_BOW.value, Achievement.OPEN_CHEST.value, Achievement.DRINK_POTION.value]

VERY_ADVANCED_ACHIEVEMENTS = [Achievement.ENTER_FIRE_REALM.value, Achievement.ENTER_ICE_REALM.value, Achievement.ENTER_GRAVEYARD.value, Achievement.DEFEAT_PIGMAN.value, Achievement.DEFEAT_FIRE_ELEMENTAL.value, Achievement.DEFEAT_FROST_TROLL.value, Achievement.DEFEAT_ICE_ELEMENTAL.value, Achievement.DAMAGE_NECROMANCER.value, Achievement.DEFEAT_NECROMANCER.value]

def achievement_mapping(achievement_value):
    if achievement_value <= 24:
        return 1
    elif achievement_value in INTERMEDIATE_ACHIEVEMENTS:
        return 3
    elif achievement_value in VERY_ADVANCED_ACHIEVEMENTS:
        return 8
    else:
        return 5

ARGS = {
    "achievement_value": 10
}
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 int, ndim 0, dtype family int, 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
apply_alpha Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
michaeltmatthews/craftax/craftax/craftax/constants.py
code served (permissive licence) · get_code("4d0be89673d8cab0")
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 ArrayImpl, ndim 3, dtype family float, shape (2, 3, 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 jax.numpy as jnp
import numpy as np

# Create a small texture tensor with shape (height, width, channels)
# The function expects shape (..., 4) where last dim is RGBA
# Using distinct sizes: 2 for height, 3 for width, 4 for channels (RGBA)
texture_data = np.random.rand(2, 3, 4).astype(np.float32)
texture = jnp.array(texture_data)

ARGS = {
    "texture": texture
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 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
add_items_from_chest Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax/game_logic.py
code served (permissive licence) · get_code("04a2736c94bfb327")
How it ran

no invocation recorded.

in_bounds Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/game_logic.py
code served (permissive licence) · get_code("69f869f3f372b3a9")
How it ran

no invocation recorded.

is_game_over Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax/game_logic.py
code served (permissive licence) · get_code("f92b11ac53caaa07")
How it ran

no invocation recorded.

is_game_over Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/game_logic.py
code served (permissive licence) · get_code("79fc8ae7f8130cd9")
How it ran

no invocation recorded.

is_in_wall Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/game_logic.py
code served (permissive licence) · get_code("e0e34adbed9cddba")
How it ran

no invocation recorded.

load_all_textures_given_size Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/constants.py
code served (permissive licence) · get_code("7bc9074fe028fcc0")
How it ran

no invocation recorded.

load_texture Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax/constants.py
code served (permissive licence) · get_code("cac576950e78d45e")
How it ran

no invocation recorded.

load_texture Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/constants.py
code served (permissive licence) · get_code("36d0610cd487a26d")
How it ran

no invocation recorded.

make_craftax_env_from_name Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_env.py
code served (permissive licence) · get_code("75146bd9aa6e42d1")
How it ran

no invocation recorded.

make_craftax_env_from_params Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_env.py
code served (permissive licence) · get_code("2b99724b4cecd006")
How it ran

no invocation recorded.

make_craftax_pixel_renderer Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/renderer.py
code served (permissive licence) · get_code("aa77320ddfdef78f")
How it ran

no invocation recorded.

render_craftax_symbolic Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/renderer.py
code served (permissive licence) · get_code("07dd74360c9c3c54")
How it ran

no invocation recorded.

render_pixels_empty Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax_classic/renderer.py
code served (permissive licence) · get_code("828f715fe8506014")
How it ran

no invocation recorded.

update_plants_with_eat Not yet run
unverified (harvested, no recorded run)
michaeltmatthews/craftax/craftax/craftax/game_logic.py
code served (permissive licence) · get_code("7bb9713c73e31d70")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 9 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 9 later) and 7 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
apply_alpha 2 other papers
2 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
2606.08340 (2026-06)
Benchmarking Open-Ended Multi-Agent Coordination in Language Agents
achievement_mapping 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
in_bounds 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
is_game_over 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
is_game_over 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
is_in_wall 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
render_craftax_symbolic 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
render_pixels_empty 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments
update_plants_with_eat 1 other paper
1 later
2506.06981 (2025-06)
Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended 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

Benchmarks play a crucial role in the development and analysis of reinforcement learning (RL) algorithms. We identify that existing benchmarks used for research into open-ended learning fall into one of two categories. Either they are too slow for meaningful research to be performed without enormous computational resources, like Crafter, NetHack and Minecraft, or they are not complex enough to pose a significant challenge, like Minigrid and Procgen. To remedy this, we first present Craftax-Classic: a ground-up rewrite of Crafter in JAX that runs up to 250x faster than the Python-native original. A run of PPO using 1 billion environment interactions finishes in under an hour using only a single GPU and averages 90% of the optimal reward. To provide a more compelling challenge we present the main Craftax benchmark, a significant extension of the Crafter mechanics with elements inspired from NetHack. Solving Craftax requires deep exploration, long term planning and memory, as well as continual adaptation to novel situations as more of the world is discovered. We show that existing methods including global and episodic exploration, as well as unsupervised environment design fail to make material progress on the benchmark. We believe that Craftax can for the first time allow researchers to experiment in a complex, open-ended environment with limited computational resources.

For agents

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

get_harvested_code_for_paper("2402.16801")
get_code_for_paper("2402.16801")
have("2402.16801")

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

curl https://syntology.ai/api/ran/2402.16801.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.16801.svg)](https://syntology.ai/paper/2402.16801)

Connect an agent — have() is free.