thomwolf

def top_k_top_p_filtering(logits, top_k=0, top_p=0.0, filter_value=-float('Inf')):
    """ Filter a distribution of logits using top-k and/or nucleus (top-p) filtering
        Args:
            logits: logits distribution shape (vocabulary size)
            top_k >0: keep only top k tokens with highest probability (top-k filtering).
            top_p >0.0: keep the top tokens with cumulative probability >= top_p (nucleus filtering).
                Nucleus filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751)
    """
    assert logits.dim() == 1  # batch size 1 for now - could be updated for more but the code would be less clear
    top_k = min(top_k, logits.size(-1))  # Safety check

ronghanghu

thomwolf

model.zero_grad()                                   # Reset gradients tensors
for i, (inputs, labels) in enumerate(training_set):
    predictions = model(inputs)                     # Forward pass
    loss = loss_function(predictions, labels)       # Compute loss function
    loss = loss / accumulation_steps                # Normalize our loss (if averaged)
    loss.backward()                                 # Backward pass
    if (i+1) % accumulation_steps == 0:             # Wait for several backward steps
        optimizer.step()                            # Now we can do an optimizer step
        model.zero_grad()                           # Reset gradients tensors
        if (i+1) % evaluation_steps == 0:           # Evaluate the model when we...

andrewjong

import torch
from torchvision import datasets

class ImageFolderWithPaths(datasets.ImageFolder):
    """Custom dataset that includes image file paths. Extends
    torchvision.datasets.ImageFolder
    """

    # override the __getitem__ method. this is the method that dataloader calls
    def __getitem__(self, index):

vadimkantorov

# References:
# [1] Multimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding, Fukui et al., https://arxiv.org/abs/1606.01847
# [2] Compact Bilinear Pooling, Gao et al., https://arxiv.org/abs/1511.06062
# [3] Fast and Scalable Polynomial Kernels via Explicit Feature Maps, Pham and Pagh, https://chbrown.github.io/kdd-2013-usb/kdd/p239.pdf
# [4] Fastfood — Approximating Kernel Expansions in Loglinear Time, Le et al., https://arxiv.org/abs/1408.3060
# [5] Original implementation in Caffe: https://github.com/gy20073/compact_bilinear_pooling

# TODO: migrate to use of new native complex64 types
# TODO: change strided x coo matmul to torch.matmul(): M[sparse_coo] @ M[strided] -> M[strided]

zhanwenchen

Updated 4/11/2018

Here's my experience of installing the NVIDIA CUDA kit 9.0 on a fresh install of Ubuntu Desktop 16.04.4 LTS.

Table of Contents

• Install NVIDIA Graphics Driver via apt-get
• Install CUDA
• Install cuDNN

Table of contents generated with markdown-toc

danielgtaylor

Moving to ES6 from CoffeeScript

I fell in love with CoffeeScript a couple of years ago. Javascript has always seemed something of an interesting curiosity to me and I was happy to see the meteoric rise of Node.js, but coming from a background of Python I really preferred a cleaner syntax.

In any fast moving community it is inevitable that things will change, and so today we see a big shift toward ES6, the new version of Javascript. It incorporates a handful of the nicer features from CoffeeScript and is usable today through tools like Babel. Here are some of my thoughts and issues on moving away from CoffeeScript in favor of ES6.

While reading I suggest keeping open a tab to Babel's learning ES6 page. The examples there are great.

Punctuation

Holy punctuation, Batman! Say goodbye to your whitespace and hello to parenthesis, curly braces, and semicolons again. Even with the advanced ES6 syntax you'll find yourself writing a lot more punctuatio

qguv

/* Solarized Dark 

For use with Jekyll and Pygments

http://ethanschoonover.com/solarized

SOLARIZED HEX      ROLE
--------- -------- ------------------------------------------
base03    #002b36  background
base01    #586e75  comments / secondary content

shlomibabluki

# coding=UTF-8
from __future__ import division
import nltk
import re
import requests

# Add your freebase key here
# If you don't have one, register at https://code.google.com/apis/console
FREEBASE_KEY = ""

nicolashery

/* Solarized Dark 

For use with Jekyll and Pygments

http://ethanschoonover.com/solarized

SOLARIZED HEX      ROLE
--------- -------- ------------------------------------------
base03    #002b36  background
base01    #586e75  comments / secondary content