Skip to content

Instantly share code, notes, and snippets.

View Shmuma's full-sized avatar

Max Lapan Shmuma

528 followers · 33 following
View GitHub Profile
@Shmuma
Shmuma / gist:0010f076035aecb3fb50460f4f6b1709
Created May 3, 2017 14:08
CartPole solution
Uses ptan library: https://github.com/Shmuma/rl/blob/ptan/ptan/samples/dqn_expreplay.py
Config for the run: https://github.com/Shmuma/rl/blob/ptan/ptan/samples/runs/dqn_exp_cartpole.ini
@Shmuma
Shmuma / gist:9b4dc5bbedc771ad44dbc82b0791bc6b
Created April 12, 2017 14:46
Crossentropy method with dense NN 40+40
https://github.com/Shmuma/Practical_RL/blob/master/week1/MountainCar-xentropy.ipynb
@Shmuma
Shmuma / gist:2781081763bfc5b19462d1432d257ca7
Created April 11, 2017 13:48
Crossentropy method
https://github.com/Shmuma/Practical_RL/blob/master/week1/taxi_crossentropy.ipynb
@Shmuma
Shmuma / gist:e370de8ecd233b4bf989e8e937e0305d
Last active April 10, 2017 13:40
Simple genetics
Genetical algorythm with mutations probability decay.
https://github.com/Shmuma/Practical_RL/blob/master/week0/frozen-8x8.ipynb
@Shmuma
Shmuma / gist:74c7b35085e7bfd5850cd0aa7f68d348
Created March 25, 2017 17:46
Async a3c for atari
https://github.com/Shmuma/rl/blob/master/algos/a3c_async.py
@Shmuma
Shmuma / Slow a3c for atari
Created March 10, 2017 08:58
Slow version: https://github.com/Shmuma/rl/blob/master/test-1/a3c_atari.py
@Shmuma
Shmuma / A3C
Created February 14, 2017 16:07
#!/usr/bin/env python
# Quick-n-dirty implementation of Advantage Actor-Critic method from https://arxiv.org/abs/1602.01783
import argparse
import logging
import numpy as np
from rl_lib.wrappers import HistoryWrapper
logger = logging.getLogger()
@Shmuma
Shmuma / cartpole-v1-474
Created February 3, 2017 22:58
#!/usr/bin/env python
# Multi-layer perceptron inspired by this: https://gym.openai.com/evaluations/eval_P4KyYPwIQdSg6EqvHgYjiw
# https://gist.githubusercontent.com/anonymous/d829ec2f8bda088ac897aa2055dcd3a8/raw/d3fcdfdcc9038bf24385589e94939dcd3c198349/crossentropy_method.py
import gym
import argparse
from gym import wrappers
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Activation
@Shmuma
Shmuma / Cartpole-v1
Created February 3, 2017 22:28
#!/usr/bin/env python
# Multi-layer perceptron inspired by this: https://gym.openai.com/evaluations/eval_P4KyYPwIQdSg6EqvHgYjiw
# https://gist.githubusercontent.com/anonymous/d829ec2f8bda088ac897aa2055dcd3a8/raw/d3fcdfdcc9038bf24385589e94939dcd3c198349/crossentropy_method.py
import gym
import argparse
from gym import wrappers
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Activation
@Shmuma
Shmuma / cartpole-v1
Created February 3, 2017 11:38
#!/usr/bin/env python
# Multi-layer perceptron inspired by this: https://gym.openai.com/evaluations/eval_P4KyYPwIQdSg6EqvHgYjiw
# https://gist.githubusercontent.com/anonymous/d829ec2f8bda088ac897aa2055dcd3a8/raw/d3fcdfdcc9038bf24385589e94939dcd3c198349/crossentropy_method.py
import gym
import argparse
from gym import wrappers
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Activation