naeval
273 строки · 7.8 Кб
1{2"cells": [3{4"cell_type": "code",5"execution_count": null,6"metadata": {},7"outputs": [],8"source": [9"%load_ext autoreload\n",10"%autoreload 2\n",11"%run -n main.py"12]13},14{15"cell_type": "markdown",16"metadata": {},17"source": [18"# datasets"19]20},21{22"cell_type": "code",23"execution_count": null,24"metadata": {},25"outputs": [],26"source": [27"# %run -n main.py\n",28"# for name in [SENT, TOKEN]:\n",29"# path = join_path(DATA_DIR, name, DATASET)\n",30"# !mkdir -p {path}"31]32},33{34"cell_type": "code",35"execution_count": null,36"metadata": {},37"outputs": [],38"source": [39"# for type in [TOKEN, SENT]:\n",40"# log(type)\n",41"# for name in [CORPORA, SYNTAG, GICRYA, RNC]:\n",42"# paths = (\n",43"# join_path(CORUS_DATA_DIR, _)\n",44"# for _ in CORUS_FILES[name]\n",45"# )\n",46"# records = (\n",47"# record\n",48"# for path in paths\n",49"# for record in DATASETS[name](path)\n",50"# )\n",51"# records = log_progress(records, desc=name)\n",52"# partitions = PARSES[type][name](records)\n",53"# lines = format_partitions(partitions)\n",54"# lines = sample(lines)\n",55"# path = join_path(DATA_DIR, type, DATASET, name + JL + GZ)\n",56"# dump_gz_lines(lines, path)"57]58},59{60"cell_type": "markdown",61"metadata": {},62"source": [63"# models"64]65},66{67"cell_type": "code",68"execution_count": null,69"metadata": {},70"outputs": [],71"source": [72"# for type in [TOKEN, SENT]:\n",73"# for name in MODELS[type]:\n",74"# path = join_path(DATA_DIR, type, name)\n",75"# !mkdir -p {path}"76]77},78{79"cell_type": "code",80"execution_count": null,81"metadata": {},82"outputs": [],83"source": [84"# !pip install rusenttokenize\n",85"# !pip install segtok\n",86"# !pip install mosestokenizer\n",87"# !pip install razdel\n",88"\n",89"# !pip install spacy\n",90"# !pip install https://github.com/aatimofeev/spacy_russian_tokenizer/archive/master.zip#egg=python-simhash\n",91"# !pip install https://github.com/Koziev/rutokenizer/archive/master.zip#egg=rutokenizer\n",92"# !pip install pymystem3\n",93"\n",94"# !pip install nltk\n",95"# nltk.download('punkt')\n",96"# !wget https://raw.githubusercontent.com/mhq/train_punkt/master/russian.pickle -O ~/nltk_data/tokenizers/punkt/PY3/russian.pickle\n",97"\n",98"# Texterra\n",99"# Можно ещё сравнивать с https://texterra.ispras.ru/products, но\n",100"# 1. она медленно работает, как минимум затраты на http\n",101"# 2. иногда кидает ошибку (возможно дело в английских предложениях)\n",102"# 3. качество немного выше segtok\n",103"\n",104"# Polyglot\n",105"# реализует http://www.unicode.org/reports/tr29/\n",106"\n",107"# Сорян, не смог установить. Дикие траблы с ICU\n",108"# brew install icu4c\n",109"# export ICU_VERSION=62.1\n",110"# export BASE=/usr/local/Cellar/icu4c/\n",111"# export PATH=$PATH:$BASE/$ICU_VERSION/bin\n",112"# export PYICU_INCLUDES=$BASE/$ICU_VERSION/include\n",113"# export PYICU_LFLAGS=-L$BASE/$ICU_VERSION/lib\n",114"# pip install pyicu polyglot\n",115"\n",116"# Вроде установилось но \n",117"# > from polyglot.text import Text\n",118"# > Text('...')\n",119"# Symbol not found: __ZNK6icu_6214Transliterator12getTargetSetERNS_10UnicodeSetE"120]121},122{123"cell_type": "code",124"execution_count": null,125"metadata": {126"scrolled": false127},128"outputs": [],129"source": [130"# for type in [SENT, TOKEN]:\n",131"# log(type)\n",132"# for model_name in MODELS[type]:\n",133"# log(model_name)\n",134"# model = MODELS[type][model_name]\n",135"# if is_class(model):\n",136"# model = model()\n",137"# model = Timing(model)\n",138"\n",139"# for dataset_name in DATASETS:\n",140"# path = join_path(DATA_DIR, type, DATASET, dataset_name + JL + GZ)\n",141"# lines = load_gz_lines(path)\n",142"# records = parse_partitions(lines)\n",143"# records = log_progress(records, desc=dataset_name)\n",144"\n",145"# records = (\n",146"# Partition.from_substrings(model(_.text))\n",147"# for _ in records\n",148"# )\n",149"# path = join_path(DATA_DIR, type, model_name, dataset_name + JL + GZ)\n",150"# lines = format_partitions(records)\n",151"# dump_gz_lines(lines, path)\n",152"\n",153"# path = join_path(DATA_DIR, STATS + JL)\n",154"# record = [[type, model_name, dataset_name], model.time]\n",155"# lines = format_jl([record])\n",156"# append_lines(lines, path)\n",157"\n",158"# model.reset()"159]160},161{162"cell_type": "markdown",163"metadata": {},164"source": [165"# score"166]167},168{169"cell_type": "code",170"execution_count": null,171"metadata": {},172"outputs": [],173"source": [174"scores = {}\n",175"keys = [\n",176" (type, model, dataset)\n",177" for type in [TOKEN, SENT]\n",178" for model in MODELS[type]\n",179" for dataset in DATASETS\n",180"]\n",181"for type, model, dataset in log_progress(keys):\n",182" path = join_path(DATA_DIR, type, DATASET, dataset + JL + GZ)\n",183" lines = load_gz_lines(path)\n",184" targets = parse_partitions(lines)\n",185"\n",186" path = join_path(DATA_DIR, type, model, dataset + JL + GZ)\n",187" lines = load_gz_lines(path)\n",188" preds = parse_partitions(lines)\n",189"\n",190" score = score_partitions(preds, targets)\n",191" scores[type, model, dataset] = score"192]193},194{195"cell_type": "markdown",196"metadata": {},197"source": [198"# report"199]200},201{202"cell_type": "code",203"execution_count": null,204"metadata": {},205"outputs": [],206"source": [207"path = join_path(DATA_DIR, STATS + JL)\n",208"lines = load_lines(path)\n",209"items = parse_jl(lines)\n",210"times = {\n",211" tuple(key): time\n",212" for key, time in items\n",213"}"214]215},216{217"cell_type": "code",218"execution_count": null,219"metadata": {},220"outputs": [],221"source": [222"token_labels = dict(model_labels(MODELS, TOKEN))\n",223"token_table = report_table(scores, times, DATASETS, MODELS, TOKEN)\n",224"html = format_report(token_table, token_labels)\n",225"patch_readme(TOKEN, html, README)\n",226"patch_readme(TOKEN, html, RAZDEL_README)\n",227"HTML(html)"228]229},230{231"cell_type": "code",232"execution_count": null,233"metadata": {},234"outputs": [],235"source": [236"sent_labels = dict(model_labels(MODELS, SENT))\n",237"sent_table = report_table(scores, times, DATASETS, MODELS, SENT)\n",238"html = format_report(sent_table, sent_labels)\n",239"patch_readme(SENT, html, README)\n",240"patch_readme(SENT, html, RAZDEL_README)\n",241"HTML(html)"242]243},244{245"cell_type": "code",246"execution_count": null,247"metadata": {},248"outputs": [],249"source": []250}251],252"metadata": {253"kernelspec": {254"display_name": "Python 3",255"language": "python",256"name": "python3"257},258"language_info": {259"codemirror_mode": {260"name": "ipython",261"version": 3262},263"file_extension": ".py",264"mimetype": "text/x-python",265"name": "python",266"nbconvert_exporter": "python",267"pygments_lexer": "ipython3",268"version": "3.6.9"269}270},271"nbformat": 4,272"nbformat_minor": 2273}274