Skip to content

Instantly share code, notes, and snippets.

View DanielTakeshi's full-sized avatar
🎯
Focusing

Daniel Seita DanielTakeshi

🎯
Focusing
View GitHub Profile
@DanielTakeshi
DanielTakeshi / tikz example
Last active January 10, 2017 01:43
An example of using tikz to generate a figure. Use `pdflatex file_name.tex` to compile, where `file_name.tex` is the name of this file. It should create an image directly in the text. Note that the package assumptions are minimal.
\documentclass[11pt]{article}
\usepackage{amsmath,amssymb,tikz}
\usetikzlibrary{positioning}
\begin{document}
\begin{center}
\begin{tikzpicture}
\node (1) [align=center] {$\mathsf{EXPSPACE}$};
\node (2) [align=center, below=3mm of 1] {$\mathsf{MA}_{\mathsf{EXP}}$};
@DanielTakeshi
DanielTakeshi / Gist for Plotting A2C
Last active May 2, 2022 23:55
For plotting some A2C results from OpenAI baselines.
"""
Quick script I ran to visualuze some A2C results. Just run `python plot.py` in the same directory
as where the stuff in `files` is located.
"""
import argparse, csv, os, pickle, sys, matplotlib
from os.path import join
matplotlib.use('Agg')
import matplotlib.pyplot as plt
plt.style.use('seaborn-darkgrid')
import numpy as np
@DanielTakeshi
DanielTakeshi / processing depth images
Last active March 19, 2020 16:44
Depth Image Processing
"""
For the BAIR Blog post.
(c) 2018 by Daniel Seita (and Michael Laskey).
"""
import numpy as np
import cv2
def depth_to_3ch(img, cutoff):
"""Useful to turn the background into black into the depth images.
import copy, cv2, os, sys, pickle, time
import numpy as np
from os.path import join
TARGET = 'tmp/'
RAW_PICKLE_FILE = 'data_raw_115_items.pkl'
def prepare_data():
"""Create the appropriate data for PyTorch using `ImageFolder`. From:
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import torchvision.models as models
from torchvision import datasets, transforms
import copy, cv2, os, sys, pickle, time
import numpy as np
from os.path import join
I am testing different PyTorch transforms on my data, and saving sample
images as they are loaded during training. See examples below.
@DanielTakeshi
DanielTakeshi / catch_test.py
Last active May 12, 2019 18:13
Exception Testing in Python
"""Some exception testing.
The code in the `try` block will stop as soon as any exception is encountered.
https://realpython.com/python-exceptions/
https://realpython.com/the-most-diabolical-python-antipattern/
"""
a = 5
@DanielTakeshi
DanielTakeshi / log_unif_example.py
Last active July 26, 2020 15:28
How to sample from a log-uniform distribution.
"""
How we might sample from a log-uniform distribution
https://stats.stackexchange.com/questions/155552/what-does-log-uniformly-distribution-mean
Only run one of these three cases at a time, otherwise the plots update each
other. Run with these versions:
matplotlib 3.2.1
numpy 1.18.3
"""
@DanielTakeshi
DanielTakeshi / gist:fec9a5cd957eb05b04b6d06a16cc88ae
Last active January 7, 2021 15:54
How to make a GIF from MuJoCo environment observations programmatically.
import gym
import numpy as np
import imageio
# Make the environment and the initial observation `o`.
env_name = 'Walker2d-v3'
env = gym.make(env_name)
obs_dim = env.observation_space.shape
act_dim = env.action_space.shape[0]
act_limit = env.action_space.high[0]
@DanielTakeshi
DanielTakeshi / gist:a4a8c431bd3ab30ed578f3b579083c7a
Last active June 16, 2021 21:56
Making subplots in matplotlib, then getting labels to appear at the bottom
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
plt.style.use('seaborn')
import numpy as np
nrows, ncols = 1, 2
fig, ax = plt.subplots(nrows, ncols, sharey=False, squeeze=True, figsize=(9*ncols, 6*nrows))
x = np.arange(100)
y0 = x + np.random.normal(loc=0.0, scale=10.0, size=100)