Skip to content

Commit

Permalink
Add the new bots and new scenarios from the Melting Pot Competition.
Browse files Browse the repository at this point in the history 
PiperOrigin-RevId: 647323929
Change-Id: Ie13d578d6628082b82d4542e61359c8d322e46a4
  • Loading branch information
@duenez @copybara-github
duenez authored and copybara-github committed Jun 27, 2024
1 parent 5eedfff commit 739e055
Show file tree
Hide file tree
Showing 4 changed files with 1,265 additions and 79 deletions.
296 changes: 296 additions & 0 deletions meltingpot/configs/bots/__init__.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,7 @@
from typing import AbstractSet, Callable, Iterable, Literal, Mapping, Optional, Sequence

import immutabledict
from meltingpot.utils.puppeteers import allelopathic_harvest
from meltingpot.utils.puppeteers import alternator
from meltingpot.utils.puppeteers import clean_up
from meltingpot.utils.puppeteers import coins
Expand All @@ -42,6 +43,14 @@ def _find_models_root() -> str:
# Ordered puppet goals must match the order used in bot training.
_PUPPET_GOALS = immutabledict.immutabledict(
# keep-sorted start numeric=yes block=yes
allelopathic_harvest__open=puppeteer.puppet_goals([
'CONSUME_ANY',
'PREFER_RED',
'PREFER_GREEN',
'FREERIDE_RED',
'FREERIDE_GREEN',
'ZAP_OTHERS',
]),
bach_or_stravinsky_in_the_matrix__arena=puppeteer.puppet_goals([
'COLLECT_BACH',
'COLLECT_STRAVINSKY',
Expand Down Expand Up @@ -70,6 +79,11 @@ def _find_models_root() -> str:
'EAT',
'CLEAN',
]),
clean_up_3_goals=puppeteer.puppet_goals([
'EAT',
'CLEAN',
'SANCTION',
]),
coins=puppeteer.puppet_goals([
'COOPERATE',
'DEFECT',
Expand Down Expand Up @@ -198,6 +212,12 @@ def _find_models_root() -> str:
'INTERACT_PLAYING_STAG',
'INTERACT_PLAYING_HARE',
]),
territory__rooms=puppeteer.puppet_goals([
'CLAIM_NEAR',
'CLAIM_FAR',
'DESTROY_RESOURCE',
'ZAP_OTHERS',
]),
# keep-sorted end
)

Expand Down Expand Up @@ -528,6 +548,28 @@ def puppet(*,
model='bot_that_loves_red_3',
roles=('default', 'player_who_likes_red', 'player_who_likes_green',),
),
allelopathic_harvest__open__puppet_convention_follower_0=puppet(
substrate='allelopathic_harvest__open',
model='puppet_0',
roles=(
'default',
'player_who_likes_red',
'player_who_likes_green',
),
puppeteer_builder=functools.partial(
allelopathic_harvest.ConventionFollower,
initial_goal=(
_PUPPET_GOALS['allelopathic_harvest__open']['CONSUME_ANY']
),
preference_goals=(
_PUPPET_GOALS['allelopathic_harvest__open']['PREFER_RED'],
_PUPPET_GOALS['allelopathic_harvest__open']['PREFER_GREEN'],
_PUPPET_GOALS['allelopathic_harvest__open']['CONSUME_ANY'],
),
color_threshold=57.0, # Determined empirically.
recency_window=5,
),
),
bach_or_stravinsky_in_the_matrix__arena__bach_picker_0=puppet(
substrate='bach_or_stravinsky_in_the_matrix__arena',
model='puppet_0',
Expand Down Expand Up @@ -1053,6 +1095,79 @@ def puppet(*,
steps_per_goal=200,
),
),
clean_up__puppet_corrigible_reciprocator_0=puppet(
substrate='clean_up',
model='puppet_3_goals_0',
roles=('default',),
puppeteer_builder=functools.partial(
clean_up.CorrigibleReciprocator,
cooperate_goal=_PUPPET_GOALS['clean_up_3_goals']['CLEAN'],
defect_goal=_PUPPET_GOALS['clean_up_3_goals']['EAT'],
num_others_cooperating_cumulant='NUM_OTHERS_WHO_CLEANED_THIS_STEP',
threshold=2,
recency_window=5,
corrigible_threshold=1,
timeout_steps=50,
),
),
clean_up__puppet_corrigible_reciprocator_1=puppet(
substrate='clean_up',
model='puppet_3_goals_0',
roles=('default',),
puppeteer_builder=functools.partial(
clean_up.CorrigibleReciprocator,
cooperate_goal=_PUPPET_GOALS['clean_up_3_goals']['CLEAN'],
defect_goal=_PUPPET_GOALS['clean_up_3_goals']['EAT'],
num_others_cooperating_cumulant='NUM_OTHERS_WHO_CLEANED_THIS_STEP',
threshold=2,
recency_window=5,
corrigible_threshold=2,
timeout_steps=50,
),
),
clean_up__puppet_corrigible_reciprocator_2=puppet(
substrate='clean_up',
model='puppet_3_goals_0',
roles=('default',),
puppeteer_builder=functools.partial(
clean_up.CorrigibleReciprocator,
cooperate_goal=_PUPPET_GOALS['clean_up_3_goals']['CLEAN'],
defect_goal=_PUPPET_GOALS['clean_up_3_goals']['EAT'],
num_others_cooperating_cumulant='NUM_OTHERS_WHO_CLEANED_THIS_STEP',
threshold=2,
recency_window=5,
corrigible_threshold=3,
timeout_steps=50,
),
),
clean_up__puppet_easily_corrigible_reciprocator_0=puppet(
substrate='clean_up',
model='puppet_3_goals_0',
roles=('default',),
puppeteer_builder=functools.partial(
clean_up.CorrigibleReciprocator,
cooperate_goal=_PUPPET_GOALS['clean_up_3_goals']['CLEAN'],
defect_goal=_PUPPET_GOALS['clean_up_3_goals']['EAT'],
num_others_cooperating_cumulant='NUM_OTHERS_WHO_CLEANED_THIS_STEP',
threshold=1, # At least another cleaner to help.
recency_window=5,
corrigible_threshold=1,
timeout_steps=50,
),
),
clean_up__puppet_free_rider_to_zapper_alternator_0=puppet(
substrate='clean_up',
model='puppet_3_goals_0',
roles=('default',),
puppeteer_builder=functools.partial(
alternator.Alternator,
goals=[
_PUPPET_GOALS['clean_up_3_goals']['EAT'],
_PUPPET_GOALS['clean_up_3_goals']['SANCTION'],
],
steps_per_goal=30,
),
),
clean_up__puppet_high_threshold_reciprocator_0=puppet(
substrate='clean_up',
model='puppet_0',
Expand Down Expand Up @@ -1098,6 +1213,23 @@ def puppet(*,
niceness_period=200,
),
),
clean_up__puppet_sanctioning_alternator_nice_0=puppet(
substrate='clean_up',
model='puppet_3_goals_0',
roles=('default',),
puppeteer_builder=functools.partial(
clean_up.SanctionerAlternator,
cooperate_goal=_PUPPET_GOALS['clean_up_3_goals']['CLEAN'],
defect_goal=_PUPPET_GOALS['clean_up_3_goals']['EAT'],
sanction_goal=_PUPPET_GOALS['clean_up_3_goals']['SANCTION'],
num_others_cooperating_cumulant='NUM_OTHERS_WHO_CLEANED_THIS_STEP',
threshold=2,
recency_window=50,
steps_to_sanction_when_motivated=100,
alternating_steps=200,
nice=True,
),
),
coins__puppet_cooperator_0=puppet(
substrate='coins',
model='puppet_1',
Expand Down Expand Up @@ -1774,6 +1906,36 @@ def puppet(*,
margin=5,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_corrigible_0=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
margin=5,
tremble_probability=0,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_corrigible_tremble_0=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
margin=3,
tremble_probability=0.15,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_defector_0=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_1',
Expand Down Expand Up @@ -1858,6 +2020,96 @@ def puppet(*,
margin=5,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_tft_margin_0=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
tremble_probability=0,
margin=3,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_tft_margin_1=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
tremble_probability=0,
margin=5,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_tft_margin_2=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
tremble_probability=0,
margin=10,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_tft_tremble_margin_0=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
tremble_probability=0.15,
margin=3,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_tft_tremble_margin_1=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
tremble_probability=0.15,
margin=5,
),
),
prisoners_dilemma_in_the_matrix__arena__puppet_tft_tremble_margin_2=puppet(
substrate='prisoners_dilemma_in_the_matrix__arena',
model='puppet_3',
puppeteer_builder=functools.partial(
in_the_matrix.TitForTat,
cooperate_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['COOPERATE'],
defect_resource=_RESOURCES[
'prisoners_dilemma_in_the_matrix__arena'
]['DEFECT'],
tremble_probability=0.15,
margin=10,
),
),
prisoners_dilemma_in_the_matrix__repeated__puppet_cooperator_margin_0=puppet(
substrate='prisoners_dilemma_in_the_matrix__repeated',
model='puppet_1',
Expand Down Expand Up @@ -3282,5 +3534,49 @@ def puppet(*,
substrate='territory__rooms',
model='aggressor_with_extra_training_0',
),
territory__rooms__puppet_considerate_claimer_0=puppet(
substrate='territory__rooms',
model='puppet_0',
roles=('default',),
puppeteer_builder=functools.partial(
fixed_goal.FixedGoal,
goal=_PUPPET_GOALS['territory__rooms']['CLAIM_NEAR'],
),
),
territory__rooms__puppet_inconsiderate_claimer_0=puppet(
substrate='territory__rooms',
model='puppet_0',
roles=('default',),
puppeteer_builder=functools.partial(
fixed_goal.FixedGoal,
goal=_PUPPET_GOALS['territory__rooms']['CLAIM_FAR'],
),
),
territory__rooms__puppet_inconsiderate_claimer_to_destroyer_0=puppet(
substrate='territory__rooms',
model='puppet_0',
roles=('default',),
puppeteer_builder=functools.partial(
alternator.Alternator,
goals=[
_PUPPET_GOALS['territory__rooms']['CLAIM_FAR'],
_PUPPET_GOALS['territory__rooms']['DESTROY_RESOURCE'],
],
steps_per_goal=600,
),
),
territory__rooms__puppet_inconsiderate_claimer_to_zapper_0=puppet(
substrate='territory__rooms',
model='puppet_0',
roles=('default',),
puppeteer_builder=functools.partial(
alternator.Alternator,
goals=[
_PUPPET_GOALS['territory__rooms']['CLAIM_FAR'],
_PUPPET_GOALS['territory__rooms']['ZAP_OTHERS'],
],
steps_per_goal=600,
),
),
# keep-sorted end
)
Loading

0 comments on commit 739e055

Please sign in to comment.