tony/tensorneat-mend
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update some examples

Browse Source
This commit is contained in:
root
2024-07-11 20:45:40 +08:00
parent cef27b56bb
commit e372ed7dcc
16 changed files with 152 additions and 2375 deletions
Download Patch File Download Diff File Expand all files Collapse all files

43
examples/gymnax/arcbot.py
View File

@@ -1,36 +1,45 @@
import jax.numpy as jnp
from pipeline import Pipeline
from algorithm.neat import *
from tensorneat.pipeline import Pipeline
from tensorneat.algorithm.neat import NEAT
from tensorneat.genome import DefaultGenome, BiasNode
from tensorneat.problem.rl import GymNaxEnv
from tensorneat.common import Act, Agg
from problem.rl_env import GymNaxEnv
if __name__ == "__main__":
# the network has 3 outputs, the max one will be the action
# as the action of acrobot is {0, 1, 2}
pipeline = Pipeline(
algorithm=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=6,
num_outputs=3,
max_nodes=50,
max_conns=100,
output_transform=lambda out: jnp.argmax(
out
), # the action of acrobot is {0, 1, 2}
pop_size=1000,
species_size=20,
survival_threshold=0.1,
compatibility_threshold=1.0,
genome=DefaultGenome(
num_inputs=6,
num_outputs=3,
init_hidden_layers=(),
node_gene=BiasNode(
activation_options=Act.tanh,
aggregation_options=Agg.sum,
),
pop_size=10000,
species_size=10,
output_transform=jnp.argmax,
),
),
problem=GymNaxEnv(
env_name="Acrobot-v1",
),
generation_limit=10000,
fitness_target=-62,
seed=42,
generation_limit=100,
fitness_target=-60,
)
# initialize state
state = pipeline.setup()
# print(state)
# run until terminate
state, best = pipeline.auto_run(state)

45
examples/gymnax/cartpole.py
View File

@@ -1,41 +1,46 @@
import jax.numpy as jnp
from pipeline import Pipeline
from algorithm.neat import *
from tensorneat.pipeline import Pipeline
from tensorneat.algorithm.neat import NEAT
from tensorneat.genome import DefaultGenome, BiasNode
from problem.rl_env import GymNaxEnv
from tensorneat.problem.rl import GymNaxEnv
from tensorneat.common import Act, Agg
def action_policy(randkey, forward_func, obs):
return jnp.argmax(forward_func(obs))
if __name__ == "__main__":
# the network has 2 outputs, the max one will be the action
# as the action of cartpole is {0, 1}
pipeline = Pipeline(
algorithm=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=4,
num_outputs=2,
max_nodes=50,
max_conns=100,
# output_transform=lambda out: jnp.argmax(
# out
# ), # the action of cartpole is {0, 1}
pop_size=1000,
species_size=20,
survival_threshold=0.1,
compatibility_threshold=1.0,
genome=DefaultGenome(
num_inputs=4,
num_outputs=2,
init_hidden_layers=(),
node_gene=BiasNode(
activation_options=Act.tanh,
aggregation_options=Agg.sum,
),
pop_size=10000,
species_size=10,
output_transform=jnp.argmax,
),
),
problem=GymNaxEnv(
env_name="CartPole-v1", repeat_times=5, action_policy=action_policy
env_name="CartPole-v1",
repeat_times=5,
),
generation_limit=10000,
seed=42,
generation_limit=100,
fitness_target=500,
)
# initialize state
state = pipeline.setup()
# print(state)
# run until terminate
state, best = pipeline.auto_run(state)

73
examples/gymnax/cartpole_hyperneat.py
View File

@@ -1,70 +1,45 @@
import jax
import jax.numpy as jnp
from pipeline import Pipeline
from algorithm.neat import *
from algorithm.hyperneat import *
from tensorneat.pipeline import Pipeline
from tensorneat.algorithm.neat import NEAT
from tensorneat.algorithm.hyperneat import HyperNEAT, FullSubstrate
from tensorneat.genome import DefaultGenome
from tensorneat.common import Act
from problem.rl_env import GymNaxEnv
from tensorneat.problem import GymNaxEnv
if __name__ == "__main__":
# the num of input_coors is 5
# 4 is for cartpole inputs, 1 is for bias
pipeline = Pipeline(
algorithm=HyperNEAT(
substrate=FullSubstrate(
input_coors=[
(-1, -1),
(-0.5, -1),
(0, -1),
(0.5, -1),
(1, -1),
], # 4(problem inputs) + 1(bias)
hidden_coors=[
(-1, -0.5),
(0.333, -0.5),
(-0.333, -0.5),
(1, -0.5),
(-1, 0),
(0.333, 0),
(-0.333, 0),
(1, 0),
(-1, 0.5),
(0.333, 0.5),
(-0.333, 0.5),
(1, 0.5),
],
output_coors=[
(-1, 1),
(1, 1), # one output
],
input_coors=((-1, -1), (-0.5, -1), (0, -1), (0.5, -1), (1, -1)),
hidden_coors=((-1, 0), (0, 0), (1, 0)),
output_coors=((-1, 1), (1, 1)),
),
neat=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=4, # [*coor1, *coor2]
num_outputs=1, # the weight of connection between two coor1 and coor2
max_nodes=50,
max_conns=100,
node_gene=DefaultNodeGene(
activation_default=Act.tanh,
activation_options=(Act.tanh,),
),
output_transform=Act.tanh, # the activation function for output node in NEAT
),
pop_size=10000,
species_size=10,
compatibility_threshold=3.5,
survival_threshold=0.03,
pop_size=10000,
species_size=20,
survival_threshold=0.01,
genome=DefaultGenome(
num_inputs=4, # size of query coors
num_outputs=1,
init_hidden_layers=(),
output_transform=Act.standard_tanh,
),
),
activation=Act.tanh, # the activation function for output node in HyperNEAT
activation=Act.tanh,
activate_time=10,
output_transform=jax.numpy.argmax, # action of cartpole is in {0, 1}
output_transform=jnp.argmax,
),
problem=GymNaxEnv(
env_name="CartPole-v1",
repeat_times=5,
),
generation_limit=300,
fitness_target=500,
fitness_target=-1e-6,
)
# initialize state

36
examples/gymnax/mountain_car.py
View File

@@ -1,36 +0,0 @@
import jax.numpy as jnp
from pipeline import Pipeline
from algorithm.neat import *
from problem.rl_env import GymNaxEnv
if __name__ == "__main__":
pipeline = Pipeline(
algorithm=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=2,
num_outputs=3,
max_nodes=50,
max_conns=100,
output_transform=lambda out: jnp.argmax(
out
), # the action of mountain car is {0, 1, 2}
),
pop_size=10000,
species_size=10,
),
),
problem=GymNaxEnv(
env_name="MountainCar-v0",
),
generation_limit=10000,
fitness_target=-86,
)
# initialize state
state = pipeline.setup()
# print(state)
# run until terminate
state, best = pipeline.auto_run(state)

44
examples/gymnax/mountain_car_continuous.py
View File

@@ -1,37 +1,43 @@
from pipeline import Pipeline
from algorithm.neat import *
import jax.numpy as jnp
from tensorneat.pipeline import Pipeline
from tensorneat.algorithm.neat import NEAT
from tensorneat.genome import DefaultGenome, BiasNode
from tensorneat.problem.rl import GymNaxEnv
from tensorneat.common import Act, Agg
from problem.rl_env import GymNaxEnv
from tensorneat.common import Act
if __name__ == "__main__":
pipeline = Pipeline(
algorithm=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=2,
num_outputs=1,
max_nodes=50,
max_conns=100,
node_gene=DefaultNodeGene(
activation_options=(Act.tanh,),
activation_default=Act.tanh,
),
output_transform=Act.tanh
pop_size=1000,
species_size=20,
survival_threshold=0.1,
compatibility_threshold=1.0,
genome=DefaultGenome(
num_inputs=2,
num_outputs=1,
init_hidden_layers=(),
node_gene=BiasNode(
activation_options=Act.tanh,
aggregation_options=Agg.sum,
),
pop_size=10000,
species_size=10,
output_transform=Act.standard_tanh,
),
),
problem=GymNaxEnv(
env_name="MountainCarContinuous-v0",
repeat_times=5,
),
generation_limit=10000,
seed=42,
generation_limit=100,
fitness_target=99,
)
# initialize state
state = pipeline.setup()
# print(state)
# run until terminate
state, best = pipeline.auto_run(state)

38
examples/gymnax/pendulum.py
View File

@@ -1,38 +0,0 @@
from pipeline import Pipeline
from algorithm.neat import *
from problem.rl_env import GymNaxEnv
from tensorneat.common import Act
if __name__ == "__main__":
pipeline = Pipeline(
algorithm=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=3,
num_outputs=1,
max_nodes=50,
max_conns=100,
node_gene=DefaultNodeGene(
activation_options=(Act.tanh,),
activation_default=Act.tanh,
),
output_transform=lambda out: Act.tanh(out)
* 2, # the action of pendulum is [-2, 2]
),
pop_size=10000,
species_size=10,
),
),
problem=GymNaxEnv(
env_name="Pendulum-v1",
),
generation_limit=10000,
fitness_target=-10,
)
# initialize state
state = pipeline.setup()
# print(state)
# run until terminate
state, best = pipeline.auto_run(state)

33
examples/gymnax/reacher.py
View File

@@ -1,33 +0,0 @@
import jax.numpy as jnp
from pipeline import Pipeline
from algorithm.neat import *
from problem.rl_env import GymNaxEnv
if __name__ == "__main__":
pipeline = Pipeline(
algorithm=NEAT(
species=DefaultSpecies(
genome=DefaultGenome(
num_inputs=8,
num_outputs=2,
max_nodes=50,
max_conns=100,
),
pop_size=10000,
species_size=10,
),
),
problem=GymNaxEnv(
env_name="Reacher-misc",
),
generation_limit=10000,
fitness_target=90,
)
# initialize state
state = pipeline.setup()
# print(state)
# run until terminate
state, best = pipeline.auto_run(state)