Skip to content

Instantly share code, notes, and snippets.

@BryanCutler

BryanCutler/tep_extending_pandas_blog1_1.ipynb

Last active May 3, 2021 16:57
Show Gist options
  • Save BryanCutler/07ab41418ba1a2575676a2dfb7ea40f9 to your computer and use it in GitHub Desktop.
Save BryanCutler/07ab41418ba1a2575676a2dfb7ea40f9 to your computer and use it in GitHub Desktop.
Download ZIP
Text Extensions for Pandas: Tips and Techniques for Extending Pandas, Part 1 Blog
Display the source blob
Display the rendered blob
Raw
tep_extending_pandas_blog1_1.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>token_id</th>\n",
" <th>span</th>\n",
" <th>input_id</th>\n",
" <th>token_type_id</th>\n",
" <th>attention_mask</th>\n",
" <th>special_tokens_mask</th>\n",
" <th>embedding</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>0</td>\n",
" <td>[0, 0): ''</td>\n",
" <td>101</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>True</td>\n",
" <td>[ -0.2746679, 0.4125931, -0.03875526...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>1</td>\n",
" <td>[0, 5): 'Monty'</td>\n",
" <td>18446</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ 0.8781024, 0.16891398, -0.0992790...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>2</td>\n",
" <td>[6, 12): 'Python'</td>\n",
" <td>18750</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ 0.44240445, 0.97607553, 0.002306254...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>3</td>\n",
" <td>[13, 16): 'and'</td>\n",
" <td>1998</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ -0.8111421, 0.61899036, -0.64434...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>4</td>\n",
" <td>[17, 20): 'the'</td>\n",
" <td>1996</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ -0.3957557, 0.51493996, -0.646472...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>5</th>\n",
" <td>5</td>\n",
" <td>[21, 25): 'Holy'</td>\n",
" <td>4151</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ 1.4014304, 0.68754816, 0.2479123...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>6</th>\n",
" <td>6</td>\n",
" <td>[26, 28): 'Gr'</td>\n",
" <td>24665</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ -0.084513076, 0.33457386, 0.2393767...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>7</th>\n",
" <td>7</td>\n",
" <td>[28, 31): 'ail'</td>\n",
" <td>12502</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>False</td>\n",
" <td>[ 0.304468, 0.19518888, -0.7640254...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>8</th>\n",
" <td>8</td>\n",
" <td>[0, 0): ''</td>\n",
" <td>102</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>True</td>\n",
" <td>[ 0.934723, 0.3263963, -0.07516271...</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" token_id span input_id token_type_id attention_mask \\\n",
"0 0 [0, 0): '' 101 0 1 \n",
"1 1 [0, 5): 'Monty' 18446 0 1 \n",
"2 2 [6, 12): 'Python' 18750 0 1 \n",
"3 3 [13, 16): 'and' 1998 0 1 \n",
"4 4 [17, 20): 'the' 1996 0 1 \n",
"5 5 [21, 25): 'Holy' 4151 0 1 \n",
"6 6 [26, 28): 'Gr' 24665 0 1 \n",
"7 7 [28, 31): 'ail' 12502 0 1 \n",
"8 8 [0, 0): '' 102 0 1 \n",
"\n",
" special_tokens_mask embedding \n",
"0 True [ -0.2746679, 0.4125931, -0.03875526... \n",
"1 False [ 0.8781024, 0.16891398, -0.0992790... \n",
"2 False [ 0.44240445, 0.97607553, 0.002306254... \n",
"3 False [ -0.8111421, 0.61899036, -0.64434... \n",
"4 False [ -0.3957557, 0.51493996, -0.646472... \n",
"5 False [ 1.4014304, 0.68754816, 0.2479123... \n",
"6 False [ -0.084513076, 0.33457386, 0.2393767... \n",
"7 False [ 0.304468, 0.19518888, -0.7640254... \n",
"8 True [ 0.934723, 0.3263963, -0.07516271... "
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"import text_extensions_for_pandas as tp\n",
"from transformers import BertTokenizerFast, BertModel\n",
"\n",
"# Create a BERT model from the Huggingface transformers library.\n",
"model_name = \"bert-base-uncased\"\n",
"tokenizer = BertTokenizerFast.from_pretrained(model_name, add_special_tokens=True)\n",
"bert = BertModel.from_pretrained(model_name)\n",
"\n",
"# Sample text to generate example extension arrays.\n",
"text = \"Monty Python and the Holy Grail\"\n",
"\n",
"# Create Pandas DataFrame with tokens and embeddings from the model.\n",
"df = tp.io.bert.make_bert_tokens(text, tokenizer)\n",
"df = tp.io.bert.add_embeddings(df, bert)\n",
"df"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"token_id int64\n",
"span SpanDtype\n",
"input_id int64\n",
"token_type_id int64\n",
"attention_mask int64\n",
"special_tokens_mask bool\n",
"embedding TensorDtype\n",
"dtype: object"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# In the above DataFrame the columns \"span\" and \"embedding\" are extension dtypes.\n",
"df.dtypes"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.6"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment