Skip to content

Instantly share code, notes, and snippets.

Quim Llimona lemonzi

Block or report user

Report or block lemonzi

Hide content and notifications from this user.

Learn more about blocking users

Contact Support about this user’s behavior.

Learn more about reporting abuse

Report abuse
View GitHub Profile
@lemonzi
lemonzi / a3c_test.py
Created Apr 24, 2017
Unit testing the A3C algorithm
View a3c_test.py
import tensorflow as tf
from tensorflow.contrib import distributions
# These are your input features, a random tensor.
# There are 128 examples per batch, 1000 time-steps, one observation.
# If your code doesn't use batches, remove the first dimension.
features = tf.random_normal([128, 1000, 1])
# Put your networks here
# policy_mu, policy_sigma = ...
@lemonzi
lemonzi / a3c.py
Created Mar 13, 2017
TensorFlow implementation of an Advantage Actor-Critic loss
View a3c.py
import numpy as np
import scipy.signal
import tensorflow as tf
import tensorify
@tensorify.tensorflow_op(tf.float32)
def discount(x, gamma=1.0, axis=0):
y = scipy.signal.lfilter([1], [1, -gamma], x[::-1], axis=axis)[::-1]
return y.astype(np.float32)
@lemonzi
lemonzi / add_name_scope.py
Last active Feb 17, 2017
@add_name_scope() decorator for TensorFlow
View add_name_scope.py
"""Decorators that make TensorFlow programming a bit easier.
Quim Llimona, 2017.
"""
import functools
import tensorflow as tf
def add_name_scope(name=None):
"""Creates a name_scope that contains all ops created by the function.
@lemonzi
lemonzi / setup.sh
Created May 24, 2014
Sphero-osc setup
View setup.sh
#/bin/bash
git clone https://github.com/lemonzi/sphero-osc
cd sphero-osc
npm install
# This is for running it
node .
You can’t perform that action at this time.