├── .gitignore └── README.md /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | pip-wheel-metadata/ 24 | share/python-wheels/ 25 | *.egg-info/ 26 | .installed.cfg 27 | *.egg 28 | MANIFEST 29 | 30 | # PyInstaller 31 | # Usually these files are written by a python script from a template 32 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 33 | *.manifest 34 | *.spec 35 | 36 | # Installer logs 37 | pip-log.txt 38 | pip-delete-this-directory.txt 39 | 40 | # Unit test / coverage reports 41 | htmlcov/ 42 | .tox/ 43 | .nox/ 44 | .coverage 45 | .coverage.* 46 | .cache 47 | nosetests.xml 48 | coverage.xml 49 | *.cover 50 | *.py,cover 51 | .hypothesis/ 52 | .pytest_cache/ 53 | 54 | # Translations 55 | *.mo 56 | *.pot 57 | 58 | # Django stuff: 59 | *.log 60 | local_settings.py 61 | db.sqlite3 62 | db.sqlite3-journal 63 | 64 | # Flask stuff: 65 | instance/ 66 | .webassets-cache 67 | 68 | # Scrapy stuff: 69 | .scrapy 70 | 71 | # Sphinx documentation 72 | docs/_build/ 73 | 74 | # PyBuilder 75 | target/ 76 | 77 | # Jupyter Notebook 78 | .ipynb_checkpoints 79 | 80 | # IPython 81 | profile_default/ 82 | ipython_config.py 83 | 84 | # pyenv 85 | .python-version 86 | 87 | # pipenv 88 | # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. 89 | # However, in case of collaboration, if having platform-specific dependencies or dependencies 90 | # having no cross-platform support, pipenv may install dependencies that don't work, or not 91 | # install all needed dependencies. 92 | #Pipfile.lock 93 | 94 | # PEP 582; used by e.g. github.com/David-OConnor/pyflow 95 | __pypackages__/ 96 | 97 | # Celery stuff 98 | celerybeat-schedule 99 | celerybeat.pid 100 | 101 | # SageMath parsed files 102 | *.sage.py 103 | 104 | # Environments 105 | .env 106 | .venv 107 | env/ 108 | venv/ 109 | ENV/ 110 | env.bak/ 111 | venv.bak/ 112 | 113 | # Spyder project settings 114 | .spyderproject 115 | .spyproject 116 | 117 | # Rope project settings 118 | .ropeproject 119 | 120 | # mkdocs documentation 121 | /site 122 | 123 | # mypy 124 | .mypy_cache/ 125 | .dmypy.json 126 | dmypy.json 127 | 128 | # Pyre type checker 129 | .pyre/ 130 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # AI Journey 2020 2 | ## создание прототипов с помощью ruGPT3 3 | 4 | [https://ai-journey.ru/contest](https://ai-journey.ru/contest) 5 | 6 | Соревнование NLP-моделей на основе ruGPT3 7 | 8 | ## Постановка задачи 9 | Необходимо разработать прототип системы, которая использует модель ruGPT3 для создания рабочего прототипа продукта. 10 | 11 | Идея оценивается по 3 параметрам: бизнес-применение, инновационность, эмоциональное вовлечение 12 | + необходим рабочий код решения 13 | 14 | Все три параметра оценивает жюри (ODS, Сбер, Альянс и тд) по шкале от 1 до 10, подробные критерии ниже. 15 | 16 | #### Критерий-1: Бизнес-применение 17 | • Суть: В критерии оценивается бизнес-потенциал проекта, его реализуемость, наличие целевой аудитории. 18 | • Пример: 19 | ◦ приложение для создания макетов сайтов по их словесному описанию https://twitter.com/jsngr/status/1287026808429383680 20 | ◦ робот-юрист переписывает жалобы и показания людей на официальный юридический язык https://twitter.com/f_j_j_/status/1283349995144359937 21 | • Шкала: от 0 до 10, где 10 – наилучший результат, 0 -наихудший. 22 | 23 | 24 | #### Критерий-2: Эмоциональное вовлечение 25 | • Суть: в этом критерии оценивается эмоциональное вовлечение пользователей в концепцию прототипа. Ваш прототип может быть основам на социально-ответственной тематике, 26 | • Пример: Стартап LearnFromAnyone создает автоматизированную систему для обучения детей предметным знаниям от первого лица: хотите изучить ракетостроение? Илон Маск расскажет как. https://learnfromanyone.com/ https://news.ycombinator.com/item?id=23870595 27 | • Шкала: от 0 до 10. 28 | 29 | 30 | #### Критерий-3: Инновации 31 | • Суть: в критерии оценивается способность предложенного проекта оптимизировать существующие процессы с помощью современных технологий, и/или улучшать качество предоставляемых услуг. 32 | • Пример: GPT-3 Tailwind CSS Code Generator (https://gpt-tailwind.com/) позволяет генерировать элементы CSS для сайта по их описанию на английском языке. Проекты такого рода 33 | • Шкала: от 0 до 10. 34 | 35 | ## Примеры 36 | 37 | Разрешается взять за основу идеи из репозитория [awesome-GPT3](https://github.com/elyase/awesome-gpt3): 38 | 39 | 40 | #### App and layout tools 41 | * [HTML layout generator](https://twitter.com/sharifshameem/status/1282676454690451457) 42 | * [Creating app design from a description](https://twitter.com/jsngr/status/1284511080715362304) 43 | * [React todo list](https://twitter.com/sharifshameem/status/1284421499915403264?s=09) 44 | * [React component based on description](https://twitter.com/sharifshameem/status/1284095222939451393) 45 | * [React component based on variable name alone](https://twitter.com/hturan/status/1282261783147958272) 46 | * [GPT-3 generating color scales from color name or emojis](https://twitter.com/hturan/status/1282381985642614790) 47 | * [Website generation in Figma from a description](https://twitter.com/jsngr/status/1287026808429383680) 48 | 49 | #### Search and data analysis 50 | * [Question answering and search engine](https://twitter.com/paraschopra/status/1284801028676653060) 51 | * [Augmenting information in tables](https://twitter.com/pavtalk/status/1285410751092416513) 52 | * [Creating charts from a description](https://twitter.com/nutanc/status/1285436266276745221) 53 | * [Natural-language interface to spreadsheet by generating code](https://twitter.com/itsyashdani/status/1285695850300219392) 54 | * [Generating and iteratively updating graphs](https://twitter.com/plotlygraphs/status/1286688715167936512) 55 | * [Guessing the movie/tv show by a description](https://www.linkedin.com/posts/mehdimabrouki_artificialintelligence-deeplearning-nlp-activity-6701068610695135232-uRur) 56 | 57 | ### Program generation and analysis 58 | * [Translating natural language into shell commmands](https://twitter.com/harlandduman/status/1282132804034150400) 59 | * [Reading code and responding to questions about it](https://twitter.com/amasad/status/1285797739930869761) 60 | * [Generating Latex from description](https://twitter.com/sh_reya/status/1284746918959239168) 61 | * [Generating SQL code 1](https://twitter.com/aquariusacquah/status/1284706786247880705) 62 | * [Generating SQL code 2](https://twitter.com/FaraazNishtar/status/1285934622891667457) 63 | * [Coding interview](https://twitter.com/lacker/status/1279136788326432771/photo/1) 64 | * [Generating python](https://twitter.com/josephbrionesaz/status/1283097878223675392) 65 | * [Generating database-specific SQL code](https://twitter.com/FaraazNishtar/status/1285934622891667457) 66 | * [AI Inceptiion: GPT-3 generating machine learning code](https://twitter.com/mattshumer_/status/1287125015528341506) 67 | * [Most Recommended Books: GPT-3 based book recommendations](http://mostrecommendedbooks.com/gpt3) 68 | 69 | 70 | #### Text generation 71 | * [Translating into several languages](https://www.johnfaben.com/blog/gpt-3-translations) 72 | * [Write this like an attorney](https://twitter.com/f_j_j_/status/1283349995144359937) 73 | * [Automatically generating Request for Admissions](https://twitter.com/f_j_j_/status/1284050844787200000) 74 | * [Writing full emails from key points](https://twitter.com/OthersideAI/status/1285776335638614017) 75 | * [Simplifying legal language](https://twitter.com/michaeltefula/status/1285505897108832257) 76 | * [Iteratively drafted non-literal poetry translation with annotations](https://imgur.com/a/3rmMVHC) 77 | * [Rephrasing sentences to be more polite](https://twitter.com/eturner303/status/1285342431244763136) 78 | * [Summarizing famous people thoughts](https://twitter.com/paraschopra/status/1284423233047900161) 79 | * [Priming GPT-3 to Speak like Any Big Five Personality](https://medium.com/intuitionmachine/priming-gpt-3-to-speak-like-any-big-five-personality-b610f5aca94f) 80 | 81 | #### Content creation 82 | * [Content creation for marketing](https://twitter.com/Siddharth87/status/1282823354567626754) 83 | * [Generating memes](https://twitter.com/wowitsmrinal/status/1287175391040290816) 84 | * [Writing Google ads](https://twitter.com/Siddharth87/status/1282823360825581568) 85 | * [Generating presentations](http://www.bemmu.com/gpt3-presentation) 86 | * [Food recipe maker](https://twitter.com/nutanc/status/1285602813385605120) 87 | * ["How to recruit board members"](https://twitter.com/zebulgar/status/1283927560435326976) 88 | * [Shakespeare-style poetry generation](https://twitter.com/Merzmensch/status/1282957710024159234) 89 | * [Generate a quiz on any topic and evaluate students answers](https://twitter.com/Learn_Awesome/status/1286189729826738176) 90 | * [Generating history questions, with answers](https://twitter.com/mckaywrigley/status/1285827683776004096) 91 | * [Text completion and style rewriting](https://twitter.com/IntuitMachine/status/1287050253103968257) 92 | 93 | 94 | ## Model 95 | ruGPT large (760M parameters) based on GPT3 paper 96 | 97 | [репозиторий с моделями](https://github.com/sberbank-ai/ru-gpts) 98 | 99 | Обучена на: 100 | - common crawl 101 | - wiki 102 | - github 103 | - stackoverflow 104 | - news 105 | - social media 106 | - books 107 | 108 | 109 | Пример работы модели - генерация сочиннения ЕГЭ по заданной теме: 110 | ``` 111 | Предмет: литература 112 | Тема: Кем является Чацкий: победителем или побежденным? (по комедии А.С. Грибоедова «Горе от ума») 113 | Сочинение: В соответствии с теми характеристиками образа, которые дает Грибоедов Чацкому в комедии «Горе от ума», он оказывается так или иначе победителем, поскольку удаляется от людей и не видит их. У Чацкого один путь — к самоубийству. Однако при этом он не производит впечатления побежденного героя. Второй путь Чацкого (эгоцентрический) — сохранить достоинство, и он не рассматривает самоубийство как путь к бессмертию. Поэтому, на наш взгляд, читателю важно понять, кто такой Чацкий и на каком пути он живет. 114 | На каком пути живет Чацкий? На его взгляд, Чацкому свойственна не только всеядность: способность говорить как человек, не чуждый эгоизму и самолюбованию. В его внутренней «темнице» развиваются внутренние конфликты, связанные с его судьбой. Чацкий хотел быть «добрым помещиком». Он хотел служить обществу, но оказался неудачником. Можно ли назвать его победителем? Нам представляется, что нет. Чацкий живет в противоречии с самим собой. Будучи членом дворянского общества, он должен служить ему, но не может соответствовать его требованиям. Человек чести должен служить своей стране, но при этом не хочет быть для общества слугой. Только вера в лучшие идеалы может помочь человеку сохранить свое достоинство. 115 | ``` 116 | 117 | 118 | ## Данные 119 | Участникам разрешается использовать любые другие текстовые данные для дообучения модели и реализации своей идеи. 120 | 121 | Дополнительные данные от организаторов: 122 | - [русская вики](https://drive.google.com/file/d/1gY6Ly_lJa4j732BItiKAW0Mfus1pFdRs/view?usp=sharing) 123 | - [github](https://drive.google.com/file/d/1GBNPYJQSdFFVsGk15W2uEx_GsXtWacyD/view?usp=sharing) 124 | - [ru_stackoverflow](https://drive.google.com/file/d/1l5pJLwRfUExahemBlqKFNmvxtA1muGz2/view?usp=sharing) 125 | - [ответы mail](https://drive.google.com/file/d/1h-P7VdSWFSd7VClEWwpexD_Fp1Zzi8qN/view?usp=sharing) 126 | - [литературные журналы](https://drive.google.com/file/d/1hU7ixLuCf40_6QwFdIPJ3YzKfRBYFwL8/view?usp=sharing) 127 | 128 | 129 | ### Формат решений 130 | 131 | Организаторам необходимо прислать: 132 | - презентацию о своей идее проекта 133 | - архив с решениями 134 | - архив с воспроизводимым кодом/Colab-ноутбук с обучением и генерацией 135 | --------------------------------------------------------------------------------