Microsoft released new version of High-Performance, Open-Source, Deep Learning Toolkit
https://news.developer.nvidia.com/microsoft-releases-new-version-of-high-performance-open-source-deep-learning-toolkit/
#microsoft #nvidia #news
https://news.developer.nvidia.com/microsoft-releases-new-version-of-high-performance-open-source-deep-learning-toolkit/
#microsoft #nvidia #news
Microsoft’s AI course is now open and free
Microsoft Professional Program for Artificial Intelligence — previously being available only to employees is now available at edx.org for free. Program includes 10 courses from basic python programming to deep learning and reinforcement learning disciplines.
https://academy.microsoft.com/en-us/professional-program/tracks/artificial-intelligence/
#mooc #microsoft #course
Microsoft Professional Program for Artificial Intelligence — previously being available only to employees is now available at edx.org for free. Program includes 10 courses from basic python programming to deep learning and reinforcement learning disciplines.
https://academy.microsoft.com/en-us/professional-program/tracks/artificial-intelligence/
#mooc #microsoft #course
Docs
Training
Master core concepts at your speed and on your schedule. Whether you've got 15 minutes or an hour, you can develop practical skills through interactive modules and paths. You can also register to learn from an instructor. Learn and grow your way.
Microsoft Research announced Open Data project! This single, cloud-hosted location offers datasets representing many years of data curation and research efforts by Microsoft.
https://www.microsoft.com/en-us/research/blog/announcing-microsoft-research-open-data-datasets-by-microsoft-research-now-available-in-the-cloud/
#data #microsoft #opensource
https://www.microsoft.com/en-us/research/blog/announcing-microsoft-research-open-data-datasets-by-microsoft-research-now-available-in-the-cloud/
#data #microsoft #opensource
Microsoft Research
Announcing Microsoft Research Open Data - Datasets by Microsoft Research now available in the cloud - Microsoft Research
The Microsoft Research Outreach team has worked extensively with the external research community to enable adoption of cloud-based research infrastructure over the past few years. Through this process, we experienced the ubiquity of Jim Gray’s fourth paradigm…
Introducing MASS – A pre-training method that outperforms BERT and GPT in sequence to sequence language generation tasks
Researchers from Microsoft Research Asia have introduced MASS—a new pre-training method that claimed to achieve better results than BERT and GPT.
Link: https://www.microsoft.com/en-us/research/blog/introducing-mass-a-pre-training-method-that-outperforms-bert-and-gpt-in-sequence-to-sequence-language-generation-tasks/
#nlp #microsoft
Researchers from Microsoft Research Asia have introduced MASS—a new pre-training method that claimed to achieve better results than BERT and GPT.
Link: https://www.microsoft.com/en-us/research/blog/introducing-mass-a-pre-training-method-that-outperforms-bert-and-gpt-in-sequence-to-sequence-language-generation-tasks/
#nlp #microsoft
Microsoft Research
Introducing MASS – A pre-training method that outperforms BERT and GPT in sequence to sequence language generation tasks
Pre-training is a hot topic in NLP research and models like BERT and GPT have definitely delivered exciting breakthroughs. The challenge is in upping our game in finer sequence to sequence based language generation tasks. Enter MASS. Click the link in our…
Microsoft open-sourced scripts and notebooks to pre-train and finetune BERT natural language model with domain-specific texts
Github: https://github.com/microsoft/AzureML-BERT
Earlier: https://yangx.top/opendatascience/837
#Bert #Microsoft #NLP #dl
Github: https://github.com/microsoft/AzureML-BERT
Earlier: https://yangx.top/opendatascience/837
#Bert #Microsoft #NLP #dl
GitHub
GitHub - microsoft/AzureML-BERT: End-to-End recipes for pre-training and fine-tuning BERT using Azure Machine Learning Service
End-to-End recipes for pre-training and fine-tuning BERT using Azure Machine Learning Service - microsoft/AzureML-BERT
Deep Learning Image Segmentation for Ecommerce Catalogue Visual Search
Microsoft’s article on image segmentation
Link: https://www.microsoft.com/developerblog/2018/04/18/deep-learning-image-segmentation-for-ecommerce-catalogue-visual-search/
#CV #DL #Segmentation #Microsoft
Microsoft’s article on image segmentation
Link: https://www.microsoft.com/developerblog/2018/04/18/deep-learning-image-segmentation-for-ecommerce-catalogue-visual-search/
#CV #DL #Segmentation #Microsoft
Open-source library provides explanation for machine learning through diverse counterfactuals
This is a development of #interpretable ML. Library to explore “what-if” scenarios for ML models.
Blog post: https://www.microsoft.com/en-us/research/blog/open-source-library-provides-explanation-for-machine-learning-through-diverse-counterfactuals/
Paper: https://www.microsoft.com/en-us/research/publication/explaining-machine-learning-classifiers-through-diverse-counterfactual-examples/
Github: https://github.com/microsoft/dice
#Microsoft #ML #opensource
This is a development of #interpretable ML. Library to explore “what-if” scenarios for ML models.
Blog post: https://www.microsoft.com/en-us/research/blog/open-source-library-provides-explanation-for-machine-learning-through-diverse-counterfactuals/
Paper: https://www.microsoft.com/en-us/research/publication/explaining-machine-learning-classifiers-through-diverse-counterfactual-examples/
Github: https://github.com/microsoft/dice
#Microsoft #ML #opensource
Microsoft Research
DiCE: Employing counterfactuals to explain machine learning algorithms
Microsoft researchers & collaborators created an open-source library to explore “what-if” scenarios for machine learning models. Learn how their method generates multiple diverse counterfactuals at once & gives insight into ML algorithm decision making.
Microsoft Research 2019 reflection—a year of progress on technology’s toughest challenges
Highlights:
* MT-DNN — a model for learning universal language embeddings that combines the multi-task learning and the language model pre-training of BERT.
* Guidelines for human-AI interaction design
* AirSim, coming from strong MS background with flight simulations, for AI realisting testing environment.
* Sand Dance, a data visualization tool included in Visual Studio Code
* Icecaps — a toolkit for conversation modeling
Link: https://www.microsoft.com/en-us/research/blog/microsoft-research-2019-reflection-a-year-of-progress-on-technologys-toughest-challenges/
#microsoft #yearinreview
Highlights:
* MT-DNN — a model for learning universal language embeddings that combines the multi-task learning and the language model pre-training of BERT.
* Guidelines for human-AI interaction design
* AirSim, coming from strong MS background with flight simulations, for AI realisting testing environment.
* Sand Dance, a data visualization tool included in Visual Studio Code
* Icecaps — a toolkit for conversation modeling
Link: https://www.microsoft.com/en-us/research/blog/microsoft-research-2019-reflection-a-year-of-progress-on-technologys-toughest-challenges/
#microsoft #yearinreview
Microsoft Research
Microsoft Research 2019 reflection—a year of progress on technology’s toughest challenges
In 2019, Microsoft researchers assembled guidelines for human-AI interaction design, explored gender bias in machine learning, and created numerous technologies that improved accessibility. Learn how these advances emphasize inclusivity.
ZeRO, DeepSpeed & Turing-NLG
ZeRO: Memory Optimization Towards Training A Trillion Parameter Models
Turing-NLG: A 17-billion-parameter language model by Microsoft
Microsoft is releasing an open-source library called DeepSpeed, which vastly advances large model training by improving scale, speed, cost, and usability, unlocking the ability to train 100-billion-parameter models; compatible with PyTorch.
ZeRO – is a new parallelized optimizer that greatly reduces the resources needed for model and data parallelism while massively increasing the number of parameters that can be trained.
ZeRO has three main optimization stages, which correspond to the partitioning of optimizer states, gradients, and parameters. When enabled cumulatively:
0. Optimizer State Partitioning (
1. Add Gradient Partitioning (
2. Add Parameter Partitioning (
They have used these breakthroughs to create Turing Natural Language Generation (Turing-NLG), the largest publicly known language model at 17 billion parameters, which you can learn more about in this accompanying blog post. Also, the abstract for Turing-NLG had been written by their own model
ZeRO & DeepSpeed: https://www.microsoft.com/en-us/research/blog/zero-deepspeed-new-system-optimizations-enable-training-models-with-over-100-billion-parameters/
paper: https://arxiv.org/abs/1910.02054
github: https://github.com/microsoft/DeepSpeed
Turing-NLG: https://www.microsoft.com/en-us/research/blog/turing-nlg-a-17-billion-parameter-language-model-by-microsoft/
#nlp #dl #ml #microsoft #deepspeed #optimization
ZeRO: Memory Optimization Towards Training A Trillion Parameter Models
Turing-NLG: A 17-billion-parameter language model by Microsoft
Microsoft is releasing an open-source library called DeepSpeed, which vastly advances large model training by improving scale, speed, cost, and usability, unlocking the ability to train 100-billion-parameter models; compatible with PyTorch.
ZeRO – is a new parallelized optimizer that greatly reduces the resources needed for model and data parallelism while massively increasing the number of parameters that can be trained.
ZeRO has three main optimization stages, which correspond to the partitioning of optimizer states, gradients, and parameters. When enabled cumulatively:
0. Optimizer State Partitioning (
P_os_) – 4x memory reduction, same communication volume as data parallelism1. Add Gradient Partitioning (
P_os+g_) – 8x memory reduction, same communication volume as data parallelism2. Add Parameter Partitioning (
P_os+g+p_) – memory reduction is linear with data parallelism degree N_d_They have used these breakthroughs to create Turing Natural Language Generation (Turing-NLG), the largest publicly known language model at 17 billion parameters, which you can learn more about in this accompanying blog post. Also, the abstract for Turing-NLG had been written by their own model
ZeRO & DeepSpeed: https://www.microsoft.com/en-us/research/blog/zero-deepspeed-new-system-optimizations-enable-training-models-with-over-100-billion-parameters/
paper: https://arxiv.org/abs/1910.02054
github: https://github.com/microsoft/DeepSpeed
Turing-NLG: https://www.microsoft.com/en-us/research/blog/turing-nlg-a-17-billion-parameter-language-model-by-microsoft/
#nlp #dl #ml #microsoft #deepspeed #optimization
Racial Disparities in Automated Speech Recognition
To no surprise, speech recognition tools have #bias due to the lack of diversity in the datasets. Group of explorers addressed that issue and provided their’s research results as a paper and #reproducible research repo.
Project link: https://fairspeech.stanford.edu
Paper: https://www.pnas.org/cgi/doi/10.1073/pnas.1915768117
Github: https://github.com/stanford-policylab/asr-disparities
#speechrecognition #voice #audiolearning #dl #microsoft #google #apple #ibm #amazon
To no surprise, speech recognition tools have #bias due to the lack of diversity in the datasets. Group of explorers addressed that issue and provided their’s research results as a paper and #reproducible research repo.
Project link: https://fairspeech.stanford.edu
Paper: https://www.pnas.org/cgi/doi/10.1073/pnas.1915768117
Github: https://github.com/stanford-policylab/asr-disparities
#speechrecognition #voice #audiolearning #dl #microsoft #google #apple #ibm #amazon
New Coding Assistant Tool From OpenAI and Microsoft
Github announced new tool for improving coding experience: Github's copilot, developed with Microsoft and OpenAI's help. This looks really promosing, at least from the announce perspective: imaging just typing convert_datetime_to_date and getting function for that. Looking forward to the actual demo.
Project: https://copilot.github.com
Blog entry: https://github.blog/2021-06-29-introducing-github-copilot-ai-pair-programmer/
CNBC news post: https://www.cnbc.com/2021/06/29/microsoft-github-copilot-ai-offers-coding-suggestions.html
#OpenAI #microsoft #coding #CS #computerlanguageunderstanding #CLU #Github
Github announced new tool for improving coding experience: Github's copilot, developed with Microsoft and OpenAI's help. This looks really promosing, at least from the announce perspective: imaging just typing convert_datetime_to_date and getting function for that. Looking forward to the actual demo.
Project: https://copilot.github.com
Blog entry: https://github.blog/2021-06-29-introducing-github-copilot-ai-pair-programmer/
CNBC news post: https://www.cnbc.com/2021/06/29/microsoft-github-copilot-ai-offers-coding-suggestions.html
#OpenAI #microsoft #coding #CS #computerlanguageunderstanding #CLU #Github
🦜 Hi!
We are the first Telegram Data Science channel.
Channel was started as a collection of notable papers, news and releases shared for the members of Open Data Science (ODS) community. Through the years of just keeping the thing going we grew to an independent online Media supporting principles of Free and Open access to the information related to Data Science.
Ultimate Posts
* Where to start learning more about Data Science. https://github.com/open-data-science/ultimate_posts/tree/master/where_to_start
* @opendatascience channel audience research. https://github.com/open-data-science/ods_channel_stats_eda
Open Data Science
ODS.ai is an international community of people anyhow related to Data Science.
Website: https://ods.ai
Hashtags
Through the years we accumulated a big collection of materials, most of them accompanied by hashtags.
#deeplearning #DL — post about deep neural networks (> 1 layer)
#cv — posts related to Computer Vision. Pictures and videos
#nlp #nlu — Natural Language Processing and Natural Language Understanding. Texts and sequences
#audiolearning #speechrecognition — related to audio information processing
#ar — augmeneted reality related content
#rl — Reinforcement Learning (agents, bots and neural networks capable of playing games)
#gan #generation #generatinveart #neuralart — about neural artt and image generation
#transformer #vqgan #vae #bert #clip #StyleGAN2 #Unet #resnet #keras #Pytorch #GPT3 #GPT2 — related to special architectures or frameworks
#coding #CS — content related to software engineering sphere
#OpenAI #microsoft #Github #DeepMind #Yandex #Google #Facebook #huggingface — hashtags related to certain companies
#productionml #sota #recommendation #embeddings #selfdriving #dataset #opensource #analytics #statistics #attention #machine #translation #visualization
Chats
- Data Science Chat https://yangx.top/datascience_chat
- ODS Slack through invite form at website
ODS resources
* Main website: https://ods.ai
* ODS Community Telegram Channel (in Russian): @ods_ru
* ML trainings Telegram Channel: @mltrainings
* ODS Community Twitter: https://twitter.com/ods_ai
Feedback and Contacts
You are welcome to reach administration through telegram bot: @opendatasciencebot
We are the first Telegram Data Science channel.
Channel was started as a collection of notable papers, news and releases shared for the members of Open Data Science (ODS) community. Through the years of just keeping the thing going we grew to an independent online Media supporting principles of Free and Open access to the information related to Data Science.
Ultimate Posts
* Where to start learning more about Data Science. https://github.com/open-data-science/ultimate_posts/tree/master/where_to_start
* @opendatascience channel audience research. https://github.com/open-data-science/ods_channel_stats_eda
Open Data Science
ODS.ai is an international community of people anyhow related to Data Science.
Website: https://ods.ai
Hashtags
Through the years we accumulated a big collection of materials, most of them accompanied by hashtags.
#deeplearning #DL — post about deep neural networks (> 1 layer)
#cv — posts related to Computer Vision. Pictures and videos
#nlp #nlu — Natural Language Processing and Natural Language Understanding. Texts and sequences
#audiolearning #speechrecognition — related to audio information processing
#ar — augmeneted reality related content
#rl — Reinforcement Learning (agents, bots and neural networks capable of playing games)
#gan #generation #generatinveart #neuralart — about neural artt and image generation
#transformer #vqgan #vae #bert #clip #StyleGAN2 #Unet #resnet #keras #Pytorch #GPT3 #GPT2 — related to special architectures or frameworks
#coding #CS — content related to software engineering sphere
#OpenAI #microsoft #Github #DeepMind #Yandex #Google #Facebook #huggingface — hashtags related to certain companies
#productionml #sota #recommendation #embeddings #selfdriving #dataset #opensource #analytics #statistics #attention #machine #translation #visualization
Chats
- Data Science Chat https://yangx.top/datascience_chat
- ODS Slack through invite form at website
ODS resources
* Main website: https://ods.ai
* ODS Community Telegram Channel (in Russian): @ods_ru
* ML trainings Telegram Channel: @mltrainings
* ODS Community Twitter: https://twitter.com/ods_ai
Feedback and Contacts
You are welcome to reach administration through telegram bot: @opendatasciencebot
GitHub
ultimate_posts/where_to_start at master · open-data-science/ultimate_posts
Ultimate posts for opendatascience telegram channel - open-data-science/ultimate_posts
👍56🔥15❤7🥰2😁2🎉2⚡1👎1👏1
Sparks of Artificial General Intelligence: Early experiments with GPT-4
TLDR: Paper from #Microsoft research about #GPT4 showing something which can be considered signs of #AGI.
ArXiV: https://arxiv.org/abs/2303.12712
TLDR: Paper from #Microsoft research about #GPT4 showing something which can be considered signs of #AGI.
ArXiV: https://arxiv.org/abs/2303.12712
arXiv.org
Sparks of Artificial General Intelligence: Early experiments with GPT-4
Artificial intelligence (AI) researchers have been developing and refining large language models (LLMs) that exhibit remarkable capabilities across a variety of domains and tasks, challenging our...
🤮17👍5❤3🔥1
Forwarded from Machinelearning
Microsoft Research обновил AutoGen Studio — Low-Code инструмент для разработчиков , предназначенный для создания, отладки и оценки многоагентных рабочих процессов.
AutoGen Studio разработан для повышения доступности среды управления локальным AI, позволяя разработчикам прототипировать и внедрять многоагентные системы без необходимости обширных знаний в области ML.
AutoGen Studio это веб-интерфейс и API Python. Он гибкий в использовании и его легко можно интегрировать его в различные среды разработки. Простой и понятный дизайн позволяет быстро собирать многоагентные системы с помощью удобного интерфейса drag-n-drop.
AutoGen Studio поддерживает API всех популярных онлайн-провейдеров LLM (OpenAI, Antрropic, Gemini, Groq, Amazon Bedrock, Corehe, MistralAI, TogetherAI ) и локальные бэкэнды :
vLLM, Ollama, LM Studio.
Возможности :
Roadmap для отслеживания новых функций, решенных проблем и запросов от сообщества разработчиков можно найти в Issues репозитория AutoGen Studio на Github.
⚠️ Примечания от разработчика:
🟠 AutoGen Studio не предназначен для использования в качестве готового к продакшену приложения. Это среда прототипирования и разработки процессов и агентов.🟠 AutoGen Studio находится в стадии активной разработки с частыми итерациями коммитов. Документация проекта обновляется синхронно с кодом.🟠 Системные требования к установке: Python 3.10+ и Node.js => 14.15.0.
@ai_machinelearning_big_data
#AI #AgentsWorkflow #MLTool #Microsoft #LLM
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
👍9🔥4❤2
Forwarded from Machinelearning
Курс содержит пошаговые инструкции с примерами кода, которые помогут научиться создавать автономных агентов с использованием машинного обучения.
Фокус на AI-агентах:
Если вас интересует именно разработка агентов — например, для симуляций, игр или интерактивных систем — данный курс будет полезен.
Каждый урок включает в себя:
- Лекцию, (видео уроки появятся в марте 2025 года)
- Примеры кода на Python с поддержкой Azure AI Foundry и Github Models
- Практические задания
- Ссылки на полезные дополнительные ресурсы
Если это ваш первый опыт работы с агентами, у Microsoft есть еще 1 курс «Генеративный ИИ для начинающих», который содержит 21 урок по построению моделей с помощью GenAI, лучше начать с него.
Переведен на 9 различных языков (русского нет).
▪ Github
@ai_machinelearning_big_data
#course #Microsoft #aiagents #ai #ml #opensource #freecourse
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥9👍6❤1
Forwarded from Machinelearning
GUI-Actor — методика на базе VLM, которая вместо традиционной генерации координат текстом при визуальной обработке интерфейса использует внимание внутри модели.
Чтобы уйти от координатного подхода, в GUI-Actor используется специальный токен
<ACTOR>, который "учится" связываться с визуальными патчами, соответствующими целевой области экрана. За один проход модель может запомнить сразу несколько кандидатов на действие.Например, все кнопки "Сохранить" в сложном интерфейсе. Это очень похоже на человеческое восприятие: видеть сам элемент, а не его позиции по осям Х и Y.
Выбрать наиболее подходящий вариант из элементов-кандидатов помогает "верификатор". Это отдельная модель, оценивающая кандидатов от
<ACTOR> и отбирающая самый подходящий для действия. Она не только улучшает точность, но и универсальна: ее можно подключить к другим моделям.Обучение требует минимум ресурсов. Можно заморозить основную VLM (Qwen2-VL-7B) и дообучить только новый action head и токены. Это всего ~100М параметров для 7B-модели.
Комбинация из такого быстрого обучения + верификатор почти догоняет полноценно обученные аналоги, сохраняя общие способности базовой модели. Никакого "катастрофического забывания" - агент учится кликать интерфейсы, не разучиваясь описывать картинки.
Результаты тестов на сложном бенчмарке ScreenSpot-Pro с высоким разрешением и незнакомыми интерфейсами (CAD, научный софт) GUI-Actor-7B с Qwen2-VL показал 40.7 балла, а с Qwen2.5-VL — 44.6, обойдя даже UI-TARS-72B (38.1).
На других тестах (ScreenSpot, ScreenSpot-v2) он тоже лидирует, особенно в иконках и текстовых элементах, демонстрируя крутую адаптацию к разным разрешениям и версткам.
В планах - выпуск еще двух моделей на основе Qwen2.5-VL (3B и 7B), демо GUI-Actor, код для модели-верификатора и датасеты для обучения.
@ai_machinelearning_big_data
#AI #ML #VLM #GUIActor #Microsoft
Please open Telegram to view this post
VIEW IN TELEGRAM
❤7👍4🔥2