Upload 2 files

convert_to_hf_dataset.ipynb

@@ -0,0 +1,2140 @@
+{
+ "cells": [
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Create HF Dataset\n",
+    "Create a huggingface dataset from the word frequencies from Danish Gigaword \n",
+    "(collected before 2022-22-01).\n",
+    "These word frequencies are from the Danish Gigaword Corpus which are tokenized using\n",
+    "the spacy pipeline for Danish `\"da_core_news_lg\"` using `spacy>=3.0.0,<3.4.0`.\n",
+    "See the script \"word_freq.py\" for more details.\n",
+    "\n"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Setup"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# !pip install datasets==2.8.0"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n"
+     ]
+    }
+   ],
+   "source": [
+    "import json\n",
+    "import os\n",
+    "from pathlib import Path\n",
+    "import pandas as pd\n",
+    "from datasets import Dataset, DatasetInfo, DatasetDict\n",
+    "import numpy as np"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# mapping from domain to top-level domain\n",
+    "domain_mapping_dict = {\n",
+    "    \"retsinformationdk\": \"Legal\",\n",
+    "    \"skat\": \"Legal\",\n",
+    "    \"retspraksis\": \"Legal\",\n",
+    "    \"hest\": \"Social Media\",\n",
+    "    \"cc\": \"Web\",\n",
+    "    \"adl\": \"Wiki & Books\",\n",
+    "    \"botxt\": \"Other\",\n",
+    "    \"danavis\": \"News\",\n",
+    "    \"dannet\": \"dannet\",\n",
+    "    \"depbank\": \"Other\",\n",
+    "    \"ep\": \"Conversation\",\n",
+    "    \"ft\": \"Conversation\",\n",
+    "    \"gutenberg\": \"Wiki & Books\",\n",
+    "    \"jvj\": \"Wiki & Books\",\n",
+    "    \"naat\": \"Conversation\",\n",
+    "    \"opensub\": \"Conversation\",\n",
+    "    \"relig\": \"Wiki & Books\",\n",
+    "    \"spont\": \"Conversation\",\n",
+    "    \"synne\": \"Other\",\n",
+    "    \"tv2r\": \"News\",\n",
+    "    \"wiki\": \"Wiki & Books\",\n",
+    "    \"wikibooks\": \"Wiki & Books\",\n",
+    "    \"wikisource\": \"Wiki & Books\",\n",
+    "    \"twfv19\": \"Social Media\",\n",
+    "}\n",
+    "\n",
+    "# mapping from domain to its long name format\n",
+    "longname_mapping_dict = {\n",
+    "    \"retsinformationdk\": \"retsinformation.dk (Danish legal information)\",\n",
+    "    \"skat\": \"Skat (Danish tax authority)\",\n",
+    "    \"retspraksis\": \"retspraksis (Danish legal information)\",\n",
+    "    \"hest\": \"Hestenettet (Danish debate forum)\",\n",
+    "    \"cc\": \"Common Crawl\",\n",
+    "    \"adl\": \" Archive for Danish Literature\",\n",
+    "    \"botxt\": \"Bornholmsk (Danish dialect)\",\n",
+    "    \"danavis\": \"Danish daily newspapers\",\n",
+    "    \"dannet\": \"DanNet (Danish WordNet)\",\n",
+    "    \"depbank\": \"Danish Dependency Treebank\",\n",
+    "    \"ep\": \"European Parliament\",\n",
+    "    \"ft\": \"Folketinget (Danish Parliament)\",\n",
+    "    \"gutenberg\": \"Gutenberg\",\n",
+    "    \"jvj\": \"Johannes V. Jensen (Danish poet)\",\n",
+    "    \"naat\": \"NAAT\",\n",
+    "    \"opensub\": \"Open Subtitles\",\n",
+    "    \"relig\": \"Religious texts\",\n",
+    "    \"spont\": \"Spontaneous speech\",\n",
+    "    \"synne\": \"Synderjysk (Danish dialect)\",\n",
+    "    \"tv2r\": \"TV 2 Radio (Danish news)\",\n",
+    "    \"wiki\": \"Wikipedia\",\n",
+    "    \"wikibooks\": \"Wikibooks\",\n",
+    "    \"wikisource\": \"Wikisource\",\n",
+    "    \"twfv19\": \"Twitter Folketingsvalget 2019 (Danish election tweets)\",\n",
+    "}\n"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## The Data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "teder {'NOUN': 1}\n",
+      "i {'ADP': 81002, 'ADV': 708, 'CCONJ': 1, 'PRON': 5, 'X': 2}\n",
+      "Marbæk {'PROPN': 3}\n",
+      "? {'PUNCT': 31558}\n",
+      "ville {'AUX': 9184, 'VERB': 177}\n",
+      "jo {'ADV': 15825, 'SCONJ': 227, 'INTJ': 88}\n",
+      "være {'AUX': 12805, 'VERB': 3773, 'X': 1, 'ADJ': 3, 'NOUN': 1, 'PRON': 1}\n"
+     ]
+    }
+   ],
+   "source": [
+    "path = Path(\"/data/DAGW/word_freqs\")\n",
+    "json_files = [str(path / f) for f in os.listdir(path) if f.endswith(\".json\")]\n",
+    "\n",
+    "\n",
+    "with open(json_files[0]) as f:\n",
+    "    data = json.load(f)\n",
+    "\n",
+    "# inspect\n",
+    "for i, (k, v) in enumerate(data.items()):\n",
+    "    print(k, v)\n",
+    "    if i > 5:\n",
+    "        break"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'dannet', 'adl', 'hest', 'wiki', 'relig', 'botxt', 'synne', 'depbank', 'cc', 'tv2r', 'opensub', 'twfv19', 'retsinformationdk', 'wikibooks', 'jvj', 'ft', 'gutenberg', 'retspraksis', 'ep', 'skat', 'spont', 'wikisource', 'naat'}\n"
+     ]
+    }
+   ],
+   "source": [
+    "def get_subdomain(json_file):\n",
+    "    return json_file.split(\"/\")[-1].split(\".\")[0].split(\"_\")[1]\n",
+    "\n",
+    "\n",
+    "domains = set([get_subdomain(json_file) for json_file in json_files])\n",
+    "print(domains)\n"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Create the Datasets\n",
+    "We will here create 4 different datasets:\n",
+    "\n",
+    "- `word_frequencies` - Danish word frequencies from Danish Gigaword.\n",
+    "- `by_domain` - word frequencies pr. domain.\n",
+    "- `with_pos ` - word frequencies pr. domain with their part-of-speech tags derived from the spacy pipeline for Danish `\"da_core_news_lg\"`.\n",
+    "- `normalized` - word frequencies pr. domain normalized by the top-level domain.\n",
+    "\n",
+    "> Note: This notebook is not very efficient, it is mainly here for documentation of the process.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def convert_to_dataset_format(json_file):\n",
+    "    with open(json_file) as f:\n",
+    "        data = json.load(f)\n",
+    "\n",
+    "    domain_origin = get_subdomain(json_file)\n",
+    "    domain = domain_mapping_dict[domain_origin]\n",
+    "\n",
+    "    return [\n",
+    "        {\n",
+    "            \"word\": word,\n",
+    "            \"pos\": pos,\n",
+    "            \"freq\": freq,\n",
+    "            \"domain\": longname_mapping_dict[domain_origin],\n",
+    "            \"domain_short\": domain_origin,\n",
+    "            \"top_level_domain\": domain,\n",
+    "        }\n",
+    "        for word, posdict in data.items()\n",
+    "        for pos, freq in posdict.items()\n",
+    "    ]\n",
+    "\n",
+    "\n",
+    "# load and convert all dataset\n",
+    "def load_and_convert_gen(n=None):\n",
+    "    for i, json_file in enumerate(json_files[:n]):\n",
+    "        print(f\"Loading {json_file} ({i+1}/{len(json_files)})\")\n",
+    "        samples = convert_to_dataset_format(json_file)\n",
+    "        for sample in samples:\n",
+    "            yield sample\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# add dataset info for each of the datasets\n",
+    "info = DatasetInfo(\n",
+    "    description=\"Danish word frequencies\",\n",
+    "    citation=\"Derczynski, L., Ciosici, M. R., Baglini, R., Christiansen, M. H., Dalsgaard, J. A., Fusaroli, R., ... & Varab, D. (2021). The Danish Gigaword Corpus. In Proceedings of the 23rd Nordic Conference on Computational Linguistics (NoDaLiDa) (pp. 413-421).\",\n",
+    "    homepage=\"https://huggingface.co/datasets/DDSC/partial-danish-gigaword-no-twitter\",\n",
+    "    version=\"1.0.0\",\n",
+    "    license=\"\",\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_11.json (1/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_12.json (2/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_1.json (3/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_12.json (4/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_2.json (5/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_46.json (6/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_24.json (7/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_20.json (8/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_18.json (9/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_relig_0.json (10/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_12.json (11/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_59.json (12/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_42.json (13/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_22.json (14/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_2.json (15/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_11.json (16/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_28.json (17/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_3.json (18/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_33.json (19/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_7.json (20/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_48.json (21/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_4.json (22/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_13.json (23/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_4.json (24/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_46.json (25/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_28.json (26/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_22.json (27/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_2.json (28/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_53.json (29/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_7.json (30/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_12.json (31/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_tv2r_4.json (32/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_10.json (33/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_tv2r_2.json (34/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_40.json (35/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_6.json (36/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_10.json (37/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_27.json (38/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_dannet_2.json (39/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_26.json (40/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_30.json (41/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_11.json (42/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_2.json (43/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_21.json (44/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_16.json (45/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_37.json (46/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_0.json (47/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_29.json (48/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_33.json (49/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_49.json (50/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_47.json (51/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_18.json (52/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_20.json (53/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_13.json (54/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_39.json (55/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_4.json (56/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_0.json (57/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_3.json (58/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_9.json (59/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_17.json (60/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_4.json (61/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_5.json (62/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_7.json (63/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_47.json (64/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_22.json (65/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_29.json (66/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_55.json (67/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_6.json (68/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_8.json (69/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_7.json (70/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_41.json (71/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_9.json (72/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_15.json (73/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_4.json (74/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_26.json (75/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_14.json (76/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_2.json (77/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_0.json (78/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_20.json (79/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_4.json (80/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_5.json (81/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_24.json (82/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_10.json (83/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_9.json (84/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_11.json (85/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_6.json (86/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_19.json (87/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_6.json (88/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_5.json (89/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_8.json (90/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_26.json (91/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_12.json (92/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_21.json (93/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_13.json (94/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_13.json (95/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_2.json (96/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_6.json (97/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_1.json (98/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_5.json (99/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_2.json (100/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_17.json (101/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_54.json (102/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_34.json (103/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_jvj_0.json (104/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_1.json (105/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_29.json (106/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_18.json (107/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_13.json (108/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_4.json (109/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_dannet_4.json (110/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_17.json (111/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_7.json (112/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_6.json (113/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_0.json (114/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_5.json (115/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_13.json (116/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_24.json (117/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_32.json (118/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_14.json (119/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_15.json (120/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_0.json (121/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_8.json (122/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_1.json (123/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_13.json (124/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_21.json (125/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_11.json (126/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_25.json (127/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_28.json (128/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_dannet_3.json (129/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_28.json (130/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_21.json (131/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_60.json (132/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_8.json (133/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_22.json (134/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_0.json (135/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_19.json (136/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_11.json (137/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_12.json (138/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_32.json (139/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_23.json (140/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_52.json (141/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_17.json (142/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_botxt_0.json (143/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_2.json (144/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_35.json (145/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_27.json (146/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_23.json (147/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_10.json (148/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_9.json (149/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_20.json (150/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_14.json (151/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_15.json (152/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_4.json (153/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_0.json (154/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_19.json (155/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_tv2r_1.json (156/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_37.json (157/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_58.json (158/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_31.json (159/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_9.json (160/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_43.json (161/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_17.json (162/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_22.json (163/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_34.json (164/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_depbank_0.json (165/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_23.json (166/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_1.json (167/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_1.json (168/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_5.json (169/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_7.json (170/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_40.json (171/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_tv2r_5.json (172/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_0.json (173/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_gutenberg_0.json (174/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_6.json (175/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_4.json (176/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_44.json (177/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_11.json (178/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_16.json (179/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_26.json (180/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_30.json (181/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_7.json (182/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_36.json (183/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_1.json (184/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_23.json (185/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_25.json (186/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_18.json (187/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_10.json (188/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_15.json (189/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_28.json (190/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_27.json (191/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_3.json (192/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_3.json (193/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wikibooks_0.json (194/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_6.json (195/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_8.json (196/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_16.json (197/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_15.json (198/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_6.json (199/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_8.json (200/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_11.json (201/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_34.json (202/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_36.json (203/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_twfv19_0.json (204/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_12.json (205/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_50.json (206/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_21.json (207/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_4.json (208/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_25.json (209/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_spont_0.json (210/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_23.json (211/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_50.json (212/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_3.json (213/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_naat_0.json (214/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_10.json (215/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_35.json (216/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wikisource_0.json (217/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_30.json (218/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_12.json (219/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_27.json (220/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_9.json (221/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_31.json (222/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_twfv19_2.json (223/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_8.json (224/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_38.json (225/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_24.json (226/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_3.json (227/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_0.json (228/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_tv2r_0.json (229/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_62.json (230/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_25.json (231/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_dannet_1.json (232/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_9.json (233/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_26.json (234/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_45.json (235/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_14.json (236/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_42.json (237/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_18.json (238/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_5.json (239/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_16.json (240/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_15.json (241/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_10.json (242/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_44.json (243/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_41.json (244/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_7.json (245/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_3.json (246/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_24.json (247/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_tv2r_3.json (248/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_3.json (249/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_56.json (250/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_8.json (251/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_16.json (252/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_5.json (253/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_cc_27.json (254/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_dannet_0.json (255/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_1.json (256/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_19.json (257/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_14.json (258/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_20.json (259/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_29.json (260/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_5.json (261/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_10.json (262/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ep_8.json (263/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_synne_0.json (264/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_57.json (265/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_2.json (266/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_51.json (267/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_61.json (268/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_twfv19_1.json (269/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_1.json (270/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_36.json (271/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_skat_2.json (272/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_25.json (273/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_3.json (274/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_48.json (275/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_49.json (276/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_ft_5.json (277/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_35.json (278/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_43.json (279/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_0.json (280/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retspraksis_6.json (281/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_45.json (282/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_37.json (283/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_9.json (284/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_7.json (285/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_opensub_31.json (286/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_39.json (287/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_14.json (288/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_3.json (289/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_38.json (290/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_retsinformationdk_33.json (291/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_wiki_19.json (292/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_hest_32.json (293/294)\n",
+      "Loading /data/DAGW/word_freqs/wordfreq_adl_1.json (294/294)\n"
+     ]
+    }
+   ],
+   "source": [
+    "# convert to huggingface dataset\n",
+    "dataset = Dataset.from_list(list(load_and_convert_gen()), info=info)\n"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Create the word_frequencies dataset"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df = dataset.to_pandas()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "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>word</th>\n",
+       "      <th>pos</th>\n",
+       "      <th>freq</th>\n",
+       "      <th>domain</th>\n",
+       "      <th>domain_short</th>\n",
+       "      <th>top_level_domain</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>teder</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>i</td>\n",
+       "      <td>ADP</td>\n",
+       "      <td>81002</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>i</td>\n",
+       "      <td>ADV</td>\n",
+       "      <td>708</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>i</td>\n",
+       "      <td>CCONJ</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>i</td>\n",
+       "      <td>PRON</td>\n",
+       "      <td>5</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035982</th>\n",
+       "      <td>Leverandører</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035983</th>\n",
+       "      <td>halvlé</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035984</th>\n",
+       "      <td>Spejlruder</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035985</th>\n",
+       "      <td>Restavrationssalen</td>\n",
+       "      <td>PROPN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035986</th>\n",
+       "      <td>Kølere</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>51035987 rows × 6 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                        word    pos   freq                             domain  \\\n",
+       "0                      teder   NOUN      1  Hestenettet (Danish debate forum)   \n",
+       "1                          i    ADP  81002  Hestenettet (Danish debate forum)   \n",
+       "2                          i    ADV    708  Hestenettet (Danish debate forum)   \n",
+       "3                          i  CCONJ      1  Hestenettet (Danish debate forum)   \n",
+       "4                          i   PRON      5  Hestenettet (Danish debate forum)   \n",
+       "...                      ...    ...    ...                                ...   \n",
+       "51035982        Leverandører   NOUN      1      Archive for Danish Literature   \n",
+       "51035983              halvlé   NOUN      1      Archive for Danish Literature   \n",
+       "51035984          Spejlruder   NOUN      1      Archive for Danish Literature   \n",
+       "51035985  Restavrationssalen  PROPN      1      Archive for Danish Literature   \n",
+       "51035986              Kølere   NOUN      1      Archive for Danish Literature   \n",
+       "\n",
+       "         domain_short top_level_domain  \n",
+       "0                hest     Social Media  \n",
+       "1                hest     Social Media  \n",
+       "2                hest     Social Media  \n",
+       "3                hest     Social Media  \n",
+       "4                hest     Social Media  \n",
+       "...               ...              ...  \n",
+       "51035982          adl     Wiki & Books  \n",
+       "51035983          adl     Wiki & Books  \n",
+       "51035984          adl     Wiki & Books  \n",
+       "51035985          adl     Wiki & Books  \n",
+       "51035986          adl     Wiki & Books  \n",
+       "\n",
+       "[51035987 rows x 6 columns]"
+      ]
+     },
+     "execution_count": 10,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/tmp/ipykernel_38251/3531764134.py:1: FutureWarning: The default value of numeric_only in DataFrameGroupBy.sum is deprecated. In a future version, numeric_only will default to False. Either specify numeric_only or select only columns which should be valid for the function.\n",
+      "  df = df.groupby([\"word\"]).sum().reset_index()\n"
+     ]
+    }
+   ],
+   "source": [
+    "df = df.groupby([\"word\"]).sum().reset_index()\n",
+    "\n",
+    "# recalculate log prob\n",
+    "total_freq = df[\"freq\"].sum()\n",
+    "df[\"log_prob\"] = np.log(df[\"freq\"] / total_freq)\n",
+    "df[\"log_prob_smoothed\"] = np.log((df[\"freq\"] + 1) / (total_freq + len(df)))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "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>word</th>\n",
+       "      <th>freq</th>\n",
+       "      <th>log_prob</th>\n",
+       "      <th>log_prob_smoothed</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>\u0001</td>\n",
+       "      <td>11523</td>\n",
+       "      <td>-11.557025</td>\n",
+       "      <td>-11.566130</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>\u0001generalforsamling</td>\n",
+       "      <td>1</td>\n",
+       "      <td>-20.909125</td>\n",
+       "      <td>-20.225169</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>\u0002</td>\n",
+       "      <td>25873</td>\n",
+       "      <td>-10.748170</td>\n",
+       "      <td>-10.757323</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>\u0003</td>\n",
+       "      <td>32510</td>\n",
+       "      <td>-10.519822</td>\n",
+       "      <td>-10.528983</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>\u0003\u0003</td>\n",
+       "      <td>7</td>\n",
+       "      <td>-18.963215</td>\n",
+       "      <td>-18.838875</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11120014</th>\n",
+       "      <td>󱤆Passer</td>\n",
+       "      <td>3</td>\n",
+       "      <td>-19.810513</td>\n",
+       "      <td>-19.532022</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11120015</th>\n",
+       "      <td>󱤈Rygning</td>\n",
+       "      <td>10</td>\n",
+       "      <td>-18.606540</td>\n",
+       "      <td>-18.520421</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11120016</th>\n",
+       "      <td>󾌵󾟛󾟛</td>\n",
+       "      <td>1</td>\n",
+       "      <td>-20.909125</td>\n",
+       "      <td>-20.225169</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11120017</th>\n",
+       "      <td>􀁸</td>\n",
+       "      <td>24</td>\n",
+       "      <td>-17.731071</td>\n",
+       "      <td>-17.699441</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11120018</th>\n",
+       "      <td>􀍴</td>\n",
+       "      <td>3</td>\n",
+       "      <td>-19.810513</td>\n",
+       "      <td>-19.532022</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>11120019 rows × 4 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                        word   freq   log_prob  log_prob_smoothed\n",
+       "0                          \u0001  11523 -11.557025         -11.566130\n",
+       "1         \u0001generalforsamling      1 -20.909125         -20.225169\n",
+       "2                          \u0002  25873 -10.748170         -10.757323\n",
+       "3                          \u0003  32510 -10.519822         -10.528983\n",
+       "4                         \u0003\u0003      7 -18.963215         -18.838875\n",
+       "...                      ...    ...        ...                ...\n",
+       "11120014             󱤆Passer      3 -19.810513         -19.532022\n",
+       "11120015            󱤈Rygning     10 -18.606540         -18.520421\n",
+       "11120016                 󾌵󾟛󾟛      1 -20.909125         -20.225169\n",
+       "11120017                   􀁸     24 -17.731071         -17.699441\n",
+       "11120018                   􀍴      3 -19.810513         -19.532022\n",
+       "\n",
+       "[11120019 rows x 4 columns]"
+      ]
+     },
+     "execution_count": 12,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df # inspect"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using custom data configuration default-db8b733866850efe\n",
+      "/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/builder.py:712: FutureWarning: 'use_auth_token' was deprecated in version 2.7.1 and will be removed in 3.0.0. Pass `use_auth_token` to the initializer/`load_dataset_builder` instead.\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Downloading and preparing dataset csv/default to /home/kenneth/.cache/huggingface/datasets/csv/default-db8b733866850efe/0.0.0...\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Downloading data files: 100%|██████████| 1/1 [00:00<00:00, 1030.29it/s]\n",
+      "Extracting data files: 100%|██████████| 1/1 [00:00<00:00, 839.70it/s]\n",
+      "Generating train split: 0 examples [00:00, ? examples/s]/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/download/streaming_download_manager.py:727: FutureWarning: the 'mangle_dupe_cols' keyword is deprecated and will be removed in a future version. Please take steps to stop the use of 'mangle_dupe_cols'\n",
+      "  return pd.read_csv(xopen(filepath_or_buffer, \"rb\", use_auth_token=use_auth_token), **kwargs)\n",
+      "                                                                       \r"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Dataset csv downloaded and prepared to /home/kenneth/.cache/huggingface/datasets/csv/default-db8b733866850efe/0.0.0. Subsequent calls will reuse this data.\n"
+     ]
+    }
+   ],
+   "source": [
+    "# save to csv\n",
+    "df.to_csv(\"danish_word_freqs.csv\", index=False)\n",
+    "\n",
+    "# load using huggingface datasets\n",
+    "word_frequencies = Dataset.from_csv(\"danish_word_freqs.csv\")"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Create the by_domain dataset\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 49,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df = dataset.to_pandas()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 50,
+   "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>word</th>\n",
+       "      <th>pos</th>\n",
+       "      <th>freq</th>\n",
+       "      <th>domain</th>\n",
+       "      <th>domain_short</th>\n",
+       "      <th>top_level_domain</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>teder</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>i</td>\n",
+       "      <td>ADP</td>\n",
+       "      <td>81002</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>i</td>\n",
+       "      <td>ADV</td>\n",
+       "      <td>708</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>i</td>\n",
+       "      <td>CCONJ</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>i</td>\n",
+       "      <td>PRON</td>\n",
+       "      <td>5</td>\n",
+       "      <td>Hestenettet (Danish debate forum)</td>\n",
+       "      <td>hest</td>\n",
+       "      <td>Social Media</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035982</th>\n",
+       "      <td>Leverandører</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035983</th>\n",
+       "      <td>halvlé</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035984</th>\n",
+       "      <td>Spejlruder</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035985</th>\n",
+       "      <td>Restavrationssalen</td>\n",
+       "      <td>PROPN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>51035986</th>\n",
+       "      <td>Kølere</td>\n",
+       "      <td>NOUN</td>\n",
+       "      <td>1</td>\n",
+       "      <td>Archive for Danish Literature</td>\n",
+       "      <td>adl</td>\n",
+       "      <td>Wiki &amp; Books</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>51035987 rows × 6 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                        word    pos   freq                             domain  \\\n",
+       "0                      teder   NOUN      1  Hestenettet (Danish debate forum)   \n",
+       "1                          i    ADP  81002  Hestenettet (Danish debate forum)   \n",
+       "2                          i    ADV    708  Hestenettet (Danish debate forum)   \n",
+       "3                          i  CCONJ      1  Hestenettet (Danish debate forum)   \n",
+       "4                          i   PRON      5  Hestenettet (Danish debate forum)   \n",
+       "...                      ...    ...    ...                                ...   \n",
+       "51035982        Leverandører   NOUN      1      Archive for Danish Literature   \n",
+       "51035983              halvlé   NOUN      1      Archive for Danish Literature   \n",
+       "51035984          Spejlruder   NOUN      1      Archive for Danish Literature   \n",
+       "51035985  Restavrationssalen  PROPN      1      Archive for Danish Literature   \n",
+       "51035986              Kølere   NOUN      1      Archive for Danish Literature   \n",
+       "\n",
+       "         domain_short top_level_domain  \n",
+       "0                hest     Social Media  \n",
+       "1                hest     Social Media  \n",
+       "2                hest     Social Media  \n",
+       "3                hest     Social Media  \n",
+       "4                hest     Social Media  \n",
+       "...               ...              ...  \n",
+       "51035982          adl     Wiki & Books  \n",
+       "51035983          adl     Wiki & Books  \n",
+       "51035984          adl     Wiki & Books  \n",
+       "51035985          adl     Wiki & Books  \n",
+       "51035986          adl     Wiki & Books  \n",
+       "\n",
+       "[51035987 rows x 6 columns]"
+      ]
+     },
+     "execution_count": 50,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 51,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/tmp/ipykernel_38251/2877160612.py:4: FutureWarning: The default value of numeric_only in DataFrameGroupBy.sum is deprecated. In a future version, numeric_only will default to False. Either specify numeric_only or select only columns which should be valid for the function.\n",
+      "  df = df.groupby([\"word\", \"domain\", \"domain_short\"]).sum().reset_index()\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "dannet\n",
+      "adl\n",
+      "hest\n",
+      "wiki\n",
+      "relig\n",
+      "botxt\n",
+      "synne\n",
+      "depbank\n",
+      "cc\n",
+      "tv2r\n",
+      "opensub\n",
+      "twfv19\n",
+      "retsinformationdk\n",
+      "wikibooks\n",
+      "jvj\n",
+      "ft\n",
+      "gutenberg\n",
+      "retspraksis\n",
+      "ep\n",
+      "skat\n",
+      "spont\n",
+      "wikisource\n",
+      "naat\n"
+     ]
+    }
+   ],
+   "source": [
+    "# hide pandas warnings\n",
+    "pd.options.mode.chained_assignment = None\n",
+    "\n",
+    "df = df.groupby([\"word\", \"domain\", \"domain_short\"]).sum().reset_index()\n",
+    "\n",
+    "# recalculate log prob\n",
+    "def calc_log_prob_pr_domain(df, domains):\n",
+    "    dfs = []\n",
+    "    for domain in domains:\n",
+    "        print(domain)\n",
+    "        df_domains = df.loc[df[\"domain_short\"] == domain]\n",
+    "        total_freq = df_domains[\"freq\"].sum()\n",
+    "        df_domains[\"log_prob\"] = np.log(df_domains[\"freq\"] / total_freq)\n",
+    "        df_domains[\"log_prob_smoothed\"] = np.log((df_domains[\"freq\"] + 1) / (total_freq + len(df_domains)))\n",
+    "        dfs.append(df_domains)\n",
+    "    \n",
+    "    return pd.concat(dfs)\n",
+    "\n",
+    "df = calc_log_prob_pr_domain(df, domains)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 52,
+   "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>word</th>\n",
+       "      <th>domain</th>\n",
+       "      <th>domain_short</th>\n",
+       "      <th>freq</th>\n",
+       "      <th>log_prob</th>\n",
+       "      <th>log_prob_smoothed</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>4486</th>\n",
+       "      <td></td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>38</td>\n",
+       "      <td>-9.948003</td>\n",
+       "      <td>-10.034464</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>14344</th>\n",
+       "      <td>!</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>686</td>\n",
+       "      <td>-7.054711</td>\n",
+       "      <td>-7.165691</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15182</th>\n",
+       "      <td>$</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>2</td>\n",
+       "      <td>-12.892442</td>\n",
+       "      <td>-12.599413</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15231</th>\n",
+       "      <td>%</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>104</td>\n",
+       "      <td>-8.941198</td>\n",
+       "      <td>-9.044065</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15272</th>\n",
+       "      <td>&amp;</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>49</td>\n",
+       "      <td>-9.693769</td>\n",
+       "      <td>-9.786003</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15308188</th>\n",
+       "      <td>’</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>123</td>\n",
+       "      <td>-7.307371</td>\n",
+       "      <td>-7.369814</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15309280</th>\n",
+       "      <td>“</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>47</td>\n",
+       "      <td>-8.269408</td>\n",
+       "      <td>-8.318895</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15309314</th>\n",
+       "      <td>”</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>139</td>\n",
+       "      <td>-7.185082</td>\n",
+       "      <td>-7.248453</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15309694</th>\n",
+       "      <td>•</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>5</td>\n",
+       "      <td>-10.510118</td>\n",
+       "      <td>-10.398336</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15310402</th>\n",
+       "      <td>…</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>11</td>\n",
+       "      <td>-9.721660</td>\n",
+       "      <td>-9.705189</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>15326084 rows × 6 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "         word                   domain domain_short  freq   log_prob  \\\n",
+       "4486           DanNet (Danish WordNet)       dannet    38  -9.948003   \n",
+       "14344       !  DanNet (Danish WordNet)       dannet   686  -7.054711   \n",
+       "15182       $  DanNet (Danish WordNet)       dannet     2 -12.892442   \n",
+       "15231       %  DanNet (Danish WordNet)       dannet   104  -8.941198   \n",
+       "15272       &  DanNet (Danish WordNet)       dannet    49  -9.693769   \n",
+       "...       ...                      ...          ...   ...        ...   \n",
+       "15308188    ’                     NAAT         naat   123  -7.307371   \n",
+       "15309280    “                     NAAT         naat    47  -8.269408   \n",
+       "15309314    ”                     NAAT         naat   139  -7.185082   \n",
+       "15309694    •                     NAAT         naat     5 -10.510118   \n",
+       "15310402    …                     NAAT         naat    11  -9.721660   \n",
+       "\n",
+       "          log_prob_smoothed  \n",
+       "4486             -10.034464  \n",
+       "14344             -7.165691  \n",
+       "15182            -12.599413  \n",
+       "15231             -9.044065  \n",
+       "15272             -9.786003  \n",
+       "...                     ...  \n",
+       "15308188          -7.369814  \n",
+       "15309280          -8.318895  \n",
+       "15309314          -7.248453  \n",
+       "15309694         -10.398336  \n",
+       "15310402          -9.705189  \n",
+       "\n",
+       "[15326084 rows x 6 columns]"
+      ]
+     },
+     "execution_count": 52,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 55,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using custom data configuration default-bcdefdf7aa1acd36\n",
+      "/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/builder.py:712: FutureWarning: 'use_auth_token' was deprecated in version 2.7.1 and will be removed in 3.0.0. Pass `use_auth_token` to the initializer/`load_dataset_builder` instead.\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Downloading and preparing dataset csv/default to /home/kenneth/.cache/huggingface/datasets/csv/default-bcdefdf7aa1acd36/0.0.0...\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Downloading data files: 100%|██████████| 1/1 [00:00<00:00, 1062.39it/s]\n",
+      "Extracting data files: 100%|██████████| 1/1 [00:00<00:00, 565.04it/s]\n",
+      "Generating train split: 0 examples [00:00, ? examples/s]/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/download/streaming_download_manager.py:727: FutureWarning: the 'mangle_dupe_cols' keyword is deprecated and will be removed in a future version. Please take steps to stop the use of 'mangle_dupe_cols'\n",
+      "  return pd.read_csv(xopen(filepath_or_buffer, \"rb\", use_auth_token=use_auth_token), **kwargs)\n",
+      "                                                                       \r"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Dataset csv downloaded and prepared to /home/kenneth/.cache/huggingface/datasets/csv/default-bcdefdf7aa1acd36/0.0.0. Subsequent calls will reuse this data.\n"
+     ]
+    }
+   ],
+   "source": [
+    "# save to csv\n",
+    "df.to_csv(\"danish_word_freqs_by_domain.csv\", index=False)\n",
+    "\n",
+    "# load using huggingface datasets\n",
+    "by_domain = Dataset.from_csv(\"danish_word_freqs_by_domain.csv\")"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Create the with_pos dataset"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 56,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/tmp/ipykernel_38251/2868179020.py:2: FutureWarning: The default value of numeric_only in DataFrameGroupBy.sum is deprecated. In a future version, numeric_only will default to False. Either specify numeric_only or select only columns which should be valid for the function.\n",
+      "  df = df.groupby([\"word\", \"domain\", \"domain_short\"]).sum().reset_index()\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "dannet\n",
+      "adl\n",
+      "hest\n",
+      "wiki\n",
+      "relig\n",
+      "botxt\n",
+      "synne\n",
+      "depbank\n",
+      "cc\n",
+      "tv2r\n",
+      "opensub\n",
+      "twfv19\n",
+      "retsinformationdk\n",
+      "wikibooks\n",
+      "jvj\n",
+      "ft\n",
+      "gutenberg\n",
+      "retspraksis\n",
+      "ep\n",
+      "skat\n",
+      "spont\n",
+      "wikisource\n",
+      "naat\n"
+     ]
+    },
+    {
+     "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>word</th>\n",
+       "      <th>domain</th>\n",
+       "      <th>domain_short</th>\n",
+       "      <th>freq</th>\n",
+       "      <th>log_prob</th>\n",
+       "      <th>log_prob_smoothed</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>4486</th>\n",
+       "      <td></td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>38</td>\n",
+       "      <td>-9.948003</td>\n",
+       "      <td>-10.034464</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>14344</th>\n",
+       "      <td>!</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>686</td>\n",
+       "      <td>-7.054711</td>\n",
+       "      <td>-7.165691</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15182</th>\n",
+       "      <td>$</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>2</td>\n",
+       "      <td>-12.892442</td>\n",
+       "      <td>-12.599413</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15231</th>\n",
+       "      <td>%</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>104</td>\n",
+       "      <td>-8.941198</td>\n",
+       "      <td>-9.044065</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15272</th>\n",
+       "      <td>&amp;</td>\n",
+       "      <td>DanNet (Danish WordNet)</td>\n",
+       "      <td>dannet</td>\n",
+       "      <td>49</td>\n",
+       "      <td>-9.693769</td>\n",
+       "      <td>-9.786003</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15308188</th>\n",
+       "      <td>’</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>123</td>\n",
+       "      <td>-7.307371</td>\n",
+       "      <td>-7.369814</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15309280</th>\n",
+       "      <td>“</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>47</td>\n",
+       "      <td>-8.269408</td>\n",
+       "      <td>-8.318895</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15309314</th>\n",
+       "      <td>”</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>139</td>\n",
+       "      <td>-7.185082</td>\n",
+       "      <td>-7.248453</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15309694</th>\n",
+       "      <td>•</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>5</td>\n",
+       "      <td>-10.510118</td>\n",
+       "      <td>-10.398336</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15310402</th>\n",
+       "      <td>…</td>\n",
+       "      <td>NAAT</td>\n",
+       "      <td>naat</td>\n",
+       "      <td>11</td>\n",
+       "      <td>-9.721660</td>\n",
+       "      <td>-9.705189</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>15326084 rows × 6 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "         word                   domain domain_short  freq   log_prob  \\\n",
+       "4486           DanNet (Danish WordNet)       dannet    38  -9.948003   \n",
+       "14344       !  DanNet (Danish WordNet)       dannet   686  -7.054711   \n",
+       "15182       $  DanNet (Danish WordNet)       dannet     2 -12.892442   \n",
+       "15231       %  DanNet (Danish WordNet)       dannet   104  -8.941198   \n",
+       "15272       &  DanNet (Danish WordNet)       dannet    49  -9.693769   \n",
+       "...       ...                      ...          ...   ...        ...   \n",
+       "15308188    ’                     NAAT         naat   123  -7.307371   \n",
+       "15309280    “                     NAAT         naat    47  -8.269408   \n",
+       "15309314    ”                     NAAT         naat   139  -7.185082   \n",
+       "15309694    •                     NAAT         naat     5 -10.510118   \n",
+       "15310402    …                     NAAT         naat    11  -9.721660   \n",
+       "\n",
+       "          log_prob_smoothed  \n",
+       "4486             -10.034464  \n",
+       "14344             -7.165691  \n",
+       "15182            -12.599413  \n",
+       "15231             -9.044065  \n",
+       "15272             -9.786003  \n",
+       "...                     ...  \n",
+       "15308188          -7.369814  \n",
+       "15309280          -8.318895  \n",
+       "15309314          -7.248453  \n",
+       "15309694         -10.398336  \n",
+       "15310402          -9.705189  \n",
+       "\n",
+       "[15326084 rows x 6 columns]"
+      ]
+     },
+     "execution_count": 56,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df = dataset.to_pandas()\n",
+    "df = df.groupby([\"word\", \"domain\", \"domain_short\"]).sum().reset_index()\n",
+    "\n",
+    "# calculate without domains log prob\n",
+    "df = calc_log_prob_pr_domain(df, domains)\n",
+    "\n",
+    "# inspect\n",
+    "df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 57,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using custom data configuration default-21900efb838e4bd8\n",
+      "/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/builder.py:712: FutureWarning: 'use_auth_token' was deprecated in version 2.7.1 and will be removed in 3.0.0. Pass `use_auth_token` to the initializer/`load_dataset_builder` instead.\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Downloading and preparing dataset csv/default to /home/kenneth/.cache/huggingface/datasets/csv/default-21900efb838e4bd8/0.0.0...\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Downloading data files: 100%|██████████| 1/1 [00:00<00:00, 1636.48it/s]\n",
+      "Extracting data files: 100%|██���███████| 1/1 [00:00<00:00, 455.46it/s]\n",
+      "Generating train split: 0 examples [00:00, ? examples/s]/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/download/streaming_download_manager.py:727: FutureWarning: the 'mangle_dupe_cols' keyword is deprecated and will be removed in a future version. Please take steps to stop the use of 'mangle_dupe_cols'\n",
+      "  return pd.read_csv(xopen(filepath_or_buffer, \"rb\", use_auth_token=use_auth_token), **kwargs)\n",
+      "                                                                       \r"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Dataset csv downloaded and prepared to /home/kenneth/.cache/huggingface/datasets/csv/default-21900efb838e4bd8/0.0.0. Subsequent calls will reuse this data.\n"
+     ]
+    }
+   ],
+   "source": [
+    "# with_pos = Dataset.from_pandas(df, info=info)\n",
+    "\n",
+    "# save to csv\n",
+    "df.to_csv(\"danish_word_freqs_with_pos.csv\", index=False)\n",
+    "with_pos = Dataset.from_csv(\"danish_word_freqs_with_pos.csv\")"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Create the normalized dataset\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 59,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/tmp/ipykernel_38251/710725097.py:6: FutureWarning: The default value of numeric_only in DataFrameGroupBy.sum is deprecated. In a future version, numeric_only will default to False. Either specify numeric_only or select only columns which should be valid for the function.\n",
+      "  df = df.groupby([\"word\", \"top_level_domain\"]).sum().reset_index()\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Social Media\n",
+      "dannet\n",
+      "Web\n",
+      "News\n",
+      "Legal\n",
+      "Wiki & Books\n",
+      "Conversation\n",
+      "Other\n",
+      "top_level_domain\n",
+      "Conversation    150531583.5\n",
+      "Legal           150531583.5\n",
+      "News            150531583.5\n",
+      "Other           150531583.5\n",
+      "Social Media    150531583.5\n",
+      "Web             150531583.5\n",
+      "Wiki & Books    150531583.5\n",
+      "dannet          150531583.5\n",
+      "Name: freq, dtype: float64\n"
+     ]
+    },
+    {
+     "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>word</th>\n",
+       "      <th>top_level_domain</th>\n",
+       "      <th>freq</th>\n",
+       "      <th>log_prob</th>\n",
+       "      <th>log_prob_smoothed</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>297</th>\n",
+       "      <td>\\n</td>\n",
+       "      <td>Social Media</td>\n",
+       "      <td>1.721349e+06</td>\n",
+       "      <td>-6.550506</td>\n",
+       "      <td>-6.561969</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>359</th>\n",
+       "      <td>\\n\\n</td>\n",
+       "      <td>Social Media</td>\n",
+       "      <td>1.643463e+03</td>\n",
+       "      <td>-13.504564</td>\n",
+       "      <td>-13.515419</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>370</th>\n",
+       "      <td>\\n\\n\\n\\n</td>\n",
+       "      <td>Social Media</td>\n",
+       "      <td>1.148070e+00</td>\n",
+       "      <td>-20.771043</td>\n",
+       "      <td>-20.156018</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>507</th>\n",
+       "      <td>\\n\\n\\n\\n \\n \\n\\n\\n \\n \\n\\n\\n \\n \\n\\n\\n \\n \\n\\n...</td>\n",
+       "      <td>Social Media</td>\n",
+       "      <td>5.740351e-01</td>\n",
+       "      <td>-21.464190</td>\n",
+       "      <td>-20.466946</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>508</th>\n",
+       "      <td>\\n\\n\\n\\n \\n \\n \\n \\n \\n \\n\\n\\n \\n \\n \\n \\n \\n ...</td>\n",
+       "      <td>Social Media</td>\n",
+       "      <td>5.740351e-01</td>\n",
+       "      <td>-21.464190</td>\n",
+       "      <td>-20.466946</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>...</th>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "      <td>...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13869006</th>\n",
+       "      <td>…</td>\n",
+       "      <td>Other</td>\n",
+       "      <td>1.269625e+04</td>\n",
+       "      <td>-11.460063</td>\n",
+       "      <td>-11.471448</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13869011</th>\n",
+       "      <td>……</td>\n",
+       "      <td>Other</td>\n",
+       "      <td>6.925227e+02</td>\n",
+       "      <td>-14.368784</td>\n",
+       "      <td>-14.378804</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13871529</th>\n",
+       "      <td>⟨hanj⟩</td>\n",
+       "      <td>Other</td>\n",
+       "      <td>4.616818e+02</td>\n",
+       "      <td>-14.774249</td>\n",
+       "      <td>-14.783549</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13871552</th>\n",
+       "      <td>⟩Gånga</td>\n",
+       "      <td>Other</td>\n",
+       "      <td>2.308409e+02</td>\n",
+       "      <td>-15.467396</td>\n",
+       "      <td>-15.474537</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13882710</th>\n",
+       "      <td>Goa</td>\n",
+       "      <td>Other</td>\n",
+       "      <td>2.308409e+02</td>\n",
+       "      <td>-15.467396</td>\n",
+       "      <td>-15.474537</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "<p>13884033 rows × 5 columns</p>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                                                       word top_level_domain  \\\n",
+       "297                                                      \\n     Social Media   \n",
+       "359                                                    \\n\\n     Social Media   \n",
+       "370                                                \\n\\n\\n\\n     Social Media   \n",
+       "507       \\n\\n\\n\\n \\n \\n\\n\\n \\n \\n\\n\\n \\n \\n\\n\\n \\n \\n\\n...     Social Media   \n",
+       "508       \\n\\n\\n\\n \\n \\n \\n \\n \\n \\n\\n\\n \\n \\n \\n \\n \\n ...     Social Media   \n",
+       "...                                                     ...              ...   \n",
+       "13869006                                                  …            Other   \n",
+       "13869011                                                 ……            Other   \n",
+       "13871529                                             ⟨hanj⟩            Other   \n",
+       "13871552                                             ⟩Gånga            Other   \n",
+       "13882710                                               Goa            Other   \n",
+       "\n",
+       "                  freq   log_prob  log_prob_smoothed  \n",
+       "297       1.721349e+06  -6.550506          -6.561969  \n",
+       "359       1.643463e+03 -13.504564         -13.515419  \n",
+       "370       1.148070e+00 -20.771043         -20.156018  \n",
+       "507       5.740351e-01 -21.464190         -20.466946  \n",
+       "508       5.740351e-01 -21.464190         -20.466946  \n",
+       "...                ...        ...                ...  \n",
+       "13869006  1.269625e+04 -11.460063         -11.471448  \n",
+       "13869011  6.925227e+02 -14.368784         -14.378804  \n",
+       "13871529  4.616818e+02 -14.774249         -14.783549  \n",
+       "13871552  2.308409e+02 -15.467396         -15.474537  \n",
+       "13882710  2.308409e+02 -15.467396         -15.474537  \n",
+       "\n",
+       "[13884033 rows x 5 columns]"
+      ]
+     },
+     "execution_count": 59,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df = dataset.to_pandas()\n",
+    "total_freq = df[\"freq\"].sum()\n",
+    "tl_domains = set(df[\"top_level_domain\"])\n",
+    "words_pr_domain = total_freq / len(tl_domains)\n",
+    "\n",
+    "df = df.groupby([\"word\", \"top_level_domain\"]).sum().reset_index()\n",
+    "\n",
+    "dfs = []\n",
+    "for dom in tl_domains:\n",
+    "    print(dom)\n",
+    "    # break\n",
+    "    dom_df = df[df[\"top_level_domain\"] == dom]\n",
+    "    dom_total_freq = dom_df[\"freq\"].sum()\n",
+    "    dom_df[\"freq\"] = dom_df[\"freq\"] * words_pr_domain / dom_total_freq\n",
+    "    assert dom_df[\"freq\"].sum() - words_pr_domain < 1e-5\n",
+    "    dfs.append(dom_df)\n",
+    "\n",
+    "df = pd.concat(dfs)\n",
+    "df[\"log_prob\"] = np.log(df[\"freq\"] / total_freq)\n",
+    "df[\"log_prob_smoothed\"] = np.log((df[\"freq\"] + 1) / (total_freq + len(df)))\n",
+    "\n",
+    "# check that the total freq is the same\n",
+    "assert df[\"freq\"].sum() - total_freq < 1e-5\n",
+    "# check that the total freq is the same pr domain\n",
+    "print(df.groupby(\"top_level_domain\")[\"freq\"].sum())\n",
+    "\n",
+    "# inspect\n",
+    "df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 60,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using custom data configuration default-dde703a9c875a737\n",
+      "/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/builder.py:712: FutureWarning: 'use_auth_token' was deprecated in version 2.7.1 and will be removed in 3.0.0. Pass `use_auth_token` to the initializer/`load_dataset_builder` instead.\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Downloading and preparing dataset csv/default to /home/kenneth/.cache/huggingface/datasets/csv/default-dde703a9c875a737/0.0.0...\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Downloading data files: 100%|██████████| 1/1 [00:00<00:00, 2280.75it/s]\n",
+      "Extracting data files: 100%|██████████| 1/1 [00:00<00:00, 569.72it/s]\n",
+      "Generating train split: 0 examples [00:00, ? examples/s]/home/kenneth/.Envs/word_freq/lib/python3.8/site-packages/datasets/download/streaming_download_manager.py:727: FutureWarning: the 'mangle_dupe_cols' keyword is deprecated and will be removed in a future version. Please take steps to stop the use of 'mangle_dupe_cols'\n",
+      "  return pd.read_csv(xopen(filepath_or_buffer, \"rb\", use_auth_token=use_auth_token), **kwargs)\n",
+      "                                                                       \r"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Dataset csv downloaded and prepared to /home/kenneth/.cache/huggingface/datasets/csv/default-dde703a9c875a737/0.0.0. Subsequent calls will reuse this data.\n"
+     ]
+    }
+   ],
+   "source": [
+    "# normalized = Dataset.from_pandas(df, info=info)\n",
+    "\n",
+    "# save to csv\n",
+    "df.to_csv(\"danish_word_freqs_normalized.csv\", index=False)\n",
+    "normalized = Dataset.from_csv(\"danish_word_freqs_normalized.csv\")"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Upload to Huggingface Hub"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 61,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "ds_dict = DatasetDict({\n",
+    "    \"word_frequencies\": word_frequencies,\n",
+    "    \"by_domain\": by_domain,\n",
+    "    \"with_pos\": with_pos,\n",
+    "    \"normalized\": normalized,\n",
+    "})"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 64,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Pushing dataset shards to the dataset hub: 100%|██████████| 1/1 [00:18<00:00, 18.75s/it]\n"
+     ]
+    }
+   ],
+   "source": [
+    "word_frequencies.push_to_hub(\"chcaa/dagw-word-frequencies\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 65,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Pushing dataset shards to the dataset hub: 100%|██████████| 3/3 [00:36<00:00, 12.04s/it]\n"
+     ]
+    }
+   ],
+   "source": [
+    "by_domain.push_to_hub(\"chcaa/dagw-word-frequencies-by-domain\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 66,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Pushing dataset shards to the dataset hub: 100%|██████████| 3/3 [00:38<00:00, 12.80s/it]\n"
+     ]
+    }
+   ],
+   "source": [
+    "with_pos.push_to_hub(\"chcaa/dagw-word-frequencies-by-domain-with-pos-tags\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 67,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Pushing dataset shards to the dataset hub: 100%|██████████| 2/2 [00:27<00:00, 13.59s/it]\n"
+     ]
+    }
+   ],
+   "source": [
+    "normalized.push_to_hub(\"chcaa/dagw-word-frequencies-normalized-by-domain\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "word_freq",
+   "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": "3.8.13"
+  },
+  "orig_nbformat": 4,
+  "vscode": {
+   "interpreter": {
+    "hash": "c10b2b53a6cd3acf03ed810a769e61720ca332994d31c08dc7197a28d260da6e"
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

wordfreq.py

@@ -0,0 +1,119 @@
+"""
+a quick script for getting wordcounts of all danish words in gigaword
+"""
+# import torch
+# import torch.multiprocessing as mp
+# mp.set_start_method('spawn', force=True)
+# torch.set_num_threads(1)
+
+import json
+import os
+from collections import Counter, defaultdict
+from pathlib import Path
+
+# from dacy.download import download_model, DEFAULT_CACHE_DIR
+from typing import List, Optional, Tuple
+
+import spacy
+
+# model = "da_dacy_large_tft-0.0.0"
+word_freq_path = "/data/DAGW/word_freqs"
+dagw_sektioner = "/data/DAGW/dagw-master/sektioner"
+
+# download_model(model, DEFAULT_CACHE_DIR)
+# path = os.path.join(DEFAULT_CACHE_DIR, model)
+
+nlp = spacy.load("da_core_news_lg", exclude=["parser", "ner"])
+# nlp.get_pipe("transformer").model.attrs["flush_cache_chance"] = 0.1
+
+
+Path(word_freq_path).mkdir(parents=True, exist_ok=True)
+
+sections = os.listdir(dagw_sektioner)
+filepaths = {}
+for p in sections:
+    subpath = os.path.join(dagw_sektioner, p)
+    filepaths[p] = [
+        os.path.join(subpath, p)
+        for p in os.listdir(subpath)
+        if p != "LICENSE" and not p.endswith(".jsonl")
+    ]
+
+
+def wordpiece_group_text(text, size=500):
+    from transformers import AutoTokenizer
+
+    tokenizer = AutoTokenizer.from_pretrained(
+        "Maltehb/-l-ctra-danish-electra-small-uncased", strip_accents=False
+    )
+    out = tokenizer.encode(text, add_special_tokens=False)
+
+    prv = 0
+    for i in range(size, len(out), size):
+        yield tokenizer.decode(out[prv:i])
+        prv = i
+    if prv < len(out):
+        yield tokenizer.decode(out[prv : len(out)])
+
+
+def group_text(text, size=2400):
+    length = len(text)
+    prv = 0
+    for i in range(size, length, size):
+        yield text[prv:i]
+        prv = i
+    if prv < length:
+        yield text[prv:length]
+
+
+def text_gen(filepaths):
+    for i, file in enumerate(filepaths):
+        if i % 10000 == 0:
+            print("\t", i, "/", len(filepaths))
+        with open(file, "r") as f:
+            text = f.read()
+            for t in group_text(text):
+                yield t
+
+
+class WordCounter:
+    def __init__(self, l: Optional[List] = None):
+        self.dict = defaultdict(lambda: defaultdict(int))
+        if l is not None:
+            self.add(l)
+
+    def add(self, l: list):
+        for token, pos in l:
+            self.dict[token][pos] += 1
+
+    def __add__(self, other):
+        for k_tok in other.dict:
+            if k_tok in self.dict:
+                for pos, count in other.dict[k_tok].items():
+                    self.dict[k_tok][pos] += count
+            else:
+                self.dict[k_tok] = other.dict[k_tok]
+        return self
+
+
+for sec in filepaths:
+    print("Starting Section:", sec)
+    docs = nlp.pipe(texts=text_gen(filepaths[sec]), n_process=10, batch_size=8)
+
+    n = 0
+    word_counts = WordCounter()
+    for i, doc in enumerate(docs, start=1):
+        word_counts += WordCounter([(t.text, t.tag_) for t in doc])
+
+        if i % 10000 == 0:
+            with open(
+                os.path.join(word_freq_path, f"wordfreq_{sec}_{n}.json"), "w"
+            ) as f:
+                json_str = json.dumps(word_counts.dict)
+                f.write(json_str)
+            word_counts = WordCounter()
+            n += 1
+
+    with open(os.path.join(word_freq_path, f"wordfreq_{sec}_{n}.json"), "w") as f:
+        json_str = json.dumps(word_counts.dict)
+        f.write(json_str)