NegatioN

#uncomment to run in a notebook
#%load_ext Cython
%%cython
cdef int round(int n, int d):
    return (n + d // 2) // d

def search(it:int, target) -> bool:
    cdef int first = 0
    cdef int last = len(target)
    cdef int middle = round(first+last, 2)

NegatioN

df = ### assumes you have a df
category_column = 'main_category'
number = 1000

dfz = []
for cat in df[category_column].unique():
   dfz.append(df[df[category_column]==cat].sample(n=number))

final_df = pd.concat(dfz)

NegatioN

torch::jit::IValue myIvalue;
torch::jit::IValue myIvalue2;
torch::jit::script::Module module;

std::unordered_map<std::string, torch::jit::IValue> umap = {{"x", myIvalue}, {"opt", myIvalue2}};
auto result = module.get_method("forward")({}, umap);


// shows all potential arguments to model forward
std::cout << module.get_method("forward").function().getSchema().arguments() << std::endl;

NegatioN

# Origin http://ubuntuhandbook.org/index.php/2020/05/canon-printer-scangear-mp-ubuntu-20-04/
# Ubuntu
sudo apt install printer-driver-gutenprint
sudo add-apt-repository ppa:thierry-f/fork-michael-gruz
sudo apt install scangearmp2
sudo apt install scangearmp-mp495series


# Arch / Manjaro

NegatioN

for f in *; do NEW_NAME=$(echo $f | sed -e "s/.*EP\([0-9][0-9]\).*/S06E\1.mkv/g"); mv ${f} ${NEW_NAME}; done;

NegatioN

\%23c	Matches in a specific column. (23), example     :%s/\%1c/#/ (insert # on first column of all rows)
:% (all rows), :. (this row), :$ (last row), :.,$ (this->last row), :5,$ (5th->last row)
s, substitute. :%s/a/b/g, change all a->b on all rows.
u, undo
p, paste
m+a, set mark a [works for a-z]
'+a, go to mark a [works for a-z]
<<, de-indent row
>>, indent row
J, join rows

NegatioN

'''
#Conda environment file (env.yml). Install with `conda env create -f env.yml`

channels:
- nvidia
- rapidsai
- anaconda
- conda-forge
dependencies:
- python=3.8

NegatioN

python -q -X faulthandler -c "#Faulty Code"

NegatioN

class MiniLMTokenizer(torch.nn.Module):
    def __init__(self, vocab: Dict[str, int], vocab_scores: Dict[str, float]):
        '''
        :param vocab: A dictionary mapping from string to index of token
        :param vocab_scores: A dictionary mapping from string to the score of a given token. This is used to decide
        which tokenization is most probable for our input string.
        For unigram models this should be avaiable under `~/.cache/torch/sentence_transformers/$YOURMODEL/unigram.json`
        You might also need to flip the scores if they're negative.
        ```
        with open(f'~/.cache/torch/sentence_transformers/{org}_{model_name}/unigram.json', 'r') as f:

NegatioN

from PIL import Image, ExifTags
from collections import defaultdict
from datetime import datetime
import os
from tqdm import tqdm
from glob import glob
import sys

glob_string = sys.argv[1]
print(f'Searching in globstring: {glob_string}')