Benjamin Klebel-Knobloch bklebel

Student for PhD in technical physics

bklebel / OxfordAdapters.py

Last active September 16, 2021 07:24

pymeasure VISA-Adapter for OxfordInstruments PS 120-10

	from pymeasure.adapters import VISAAdapter
	from pyvisa.errors import VisaIOError

	import re
	import logging


	log = logging.getLogger(__name__)
	log.addHandler(logging.NullHandler())

bklebel / processors.py

Last active August 15, 2018 08:39

MultiInputProcessor for keras-rl without window_length

	import numpy as np

	from rl.core import Processor
	from rl.util import WhiteningNormalizer

	class MultiInputProcessor(Processor):
	"""Converts observations from an environment with multiple observations for use in a neural network
	policy.
	In some cases, you have environments that return multiple different observations per timestep
	(in a robotics context, for example, a camera may be used to view the scene and a joint encoder may

bklebel / cartpole_multidict.py

Last active October 14, 2019 15:59

Cartpole Example for dictionary observation spaces, using keras-rl


	import math
	import gym
	from gym import spaces, logger
	from gym.utils import seeding
	import numpy as np


	class CartPoleMultidictEnv(gym.Env):
	metadata = {

bklebel / MultiInputProcessor.py

Last active September 6, 2019 09:29

Keras model using multiple separate inputs to separate networks - working with keras-rl


	class MultiInputProcessor(Processor):
	"""Converts observations from an environment with multiple observations for use in a neural network
	policy.
	In some cases, you have environments that return multiple different observations per timestep
	(in a robotics context, for example, a camera may be used to view the scene and a joint encoder may
	be used to report the angles for each joint). Usually, this can be handled by a policy that has
	multiple inputs, one for each modality. However, observations are returned by the environment
	in the form of a tuple `[(modality1_t, modality2_t, ..., modalityn_t) for t in T]` but the neural network
	expects them in per-modality batches like so: `[[modality1_1, ..., modality1_T], ..., [[modalityn_1, ..., modalityn_T]]`.