johnhw

import numpy as np

def softmax(a,b):
    return np.log(np.exp(a)+np.exp(b))

def softmin(a,b):
    return -softmax(-a, -b)

def softrank(a, b):
    return softmin(a,b), softmax(a,b)

johnhw

from tqdm.auto import trange, tqdm
def tqdm_train_loop(train_fn, num_epochs, train_loader, val_fn=None):
    # each epoch
    with trange(num_epochs, unit="epoch") as epoch_t:
        for epoch in epoch_t:
            # each batch
            with tqdm(train_loader, leave=False, unit='batch', postfix='') as batch_t:
                for batch_idx, (features, targets) in enumerate(batch_t):
                    cost = train_fn(features, labels)
                    batch_t.postfix = f"cost {cost.item():.2f}"

johnhw

import numpy as np
import matplotlib as mpl
import matplotlib.pyplot as plt
from collections import defaultdict
import colorsys
import munkres  # provides Hungarian algorithm
import scipy.spatial.distance as dist
import functools
import os

johnhw

johnhw

# This implements the ideas in the paper "Neural Ordinary Differential Equations"
# in as simple a form as possible, using only autograd. It is not efficient.
# It is not useful for any practical purpose. 
# Use [torchdiffeq](https://github.com/rtqichen/torchdiffeq) for any real use.
#
# > [1] Ricky T. Q. Chen, Yulia Rubanova, Jesse Bettencourt, David Duvenaud. 
# "Neural Ordinary Differential Equations." *Advances in Neural Processing Information Systems.* 2018. 
# [[arxiv]](https://arxiv.org/abs/1806.07366)
#
# The implementation is based on the 

johnhw

johnhw

A complete list of books, articles, blog posts, videos and neat pages that support Data Fundamentals (H), organised by Unit.

Formatting

If the resource is available online (legally) I have included a link to it. Each entry has symbols following it.

• ⨕⨕⨕ indicates difficulty/depth, from ⨕ (easy to pick up intro, no background required) through ⨕⨕⨕⨕⨕ (graduate level textbook, maths heavy, expect equations)
• ⭐ indicates a particularly recommended resource; 🌟 is a very strongly recommended resource and you should look at it.

johnhw

# BSD 3-Clause License

# Copyright (c) 2018, John H. Williamson
# All rights reserved.

# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:

# * Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.

johnhw

$$\newcommand{\vec}[1]{{\bf #1} } 
\newenvironment{examinable}{\hspace{16em}}{{\hspace{16em}\LARGE[\spadesuit]}}
\newcommand{\real}{\mathbb{R}}
\newcommand{\expect}[1]{\mathbb{E}[#1]}
\DeclareMathOperator*{\argmin}{arg\,min}
\begin{examinable}
\vec{x}
\real
\end{examinable}
$$

johnhw

-- assign logical name to macro keyboard
lmc_assign_keyboard('MACROS');

view_mode = false
-- define callback for whole device
lmc_set_handler('MACROS',function(button, direction)

print(button)
  if (direction == 1) then return end  -- ignore down
  if     (button == 33) then lmc_send_keys("{ESC}{UP}")