Reinforcement learning – AI.RadicalShift

researchRevealing the Mind of a Language Model: Pythia and the Transparent Science of LLMs

Americas, Artificial Intelligence, Generative AI, Internet & Online, Language model training, LLMs (Large Language Models), Machine learning, Natural language processing (NLP), Reinforcement learning, Retrieval-augmented generation (RAG), Science, SLMs (Small Language Models)

May 2, 2025

Oleg Lazarov

300 billion tokens, AI Ethics, AI Fairness, Apache-2.0, autoregressive transformer models, bias, bias mitigation, BigScience BLOOM, BigScience’s BLOOM, BLOOM, data deduplication, debugging, dynamics of training, EleutherAI, ethics, GitHub, GPT-3, GPT-4, GPT-J, GPT-Neo, HELM, HELM benchmark, Holistic Evaluation of Language Models, Hugging Face, interpretability, K2, layered transparency, LLM360’s Amber, LLMs, memorization, Meta’s OPT, MinHash-LSH, ML researcher, NLP Benchmarking, OLMo, Open Language Model, Open LLM Leaderboard, OpenAI, PaLM, Poisson Point Process, PolyPythias, public text corpus, Pythia, researchers, scaling laws, Stanford, Stanford CRFM, Stella Biderman, term frequency effects, The Pile, the ROOTS corpus, training data order, transparency in training, trustworthy AI

Large language models (LLMs) have achieved remarkable capabilities in recent years, but how they learn and evolve during training remains something of a mystery. In response, EleutherAI’s Pythia project takes an unprecedented approach: instead of releasing just one big model, Pythia provides an entire suite of 16 language models (ranging from 70 million to 12…
Continue Reading
newsBoston Dynamics Teams Up with Robotics & AI Institute to Advance Reinforcement Learning for Atlas Humanoid Robot

Americas, Artificial Intelligence, Europe, Industrial, Reinforcement learning, Robotics

February 6, 2025

Oleg Lazarov

AI applications, AI development, AI robotics, Atlas robot, Boston Dynamics, Dynamic Behavior, Full-Body Contact, humanoid robot development, humanoid robots, Loco-Manipulation, Marc Raibert, Mobile Manipulation, reinforcement learning, Robert Playter, robot manipulation, robot mobility, Robotics & AI Institute, robotics research, Sim-to-Real

BOSTON/ CAMBRIDGE, MA — In a major step for humanoid robotics, Boston Dynamics and the Robotics & AI Institute (RAI Institute) have come together to push the boundaries of reinforcement learning and its application to the Atlas robot. This partnership seeks to establish a shared training pipeline that will enable the new electric Atlas robot…
Continue Reading
newsNVIDIA Accelerates Physical AI with Advanced Platforms for Humanoid Robots and Autonomous Facilities

AI assistants, Americas, Artificial Intelligence, Automation, Digital twins, Edge computing, Generative AI, Industrial, Reinforcement learning, Robotics, Transportation & Logistics

October 23, 2024

Oleg Lazarov

Agility Robotics, Amazon Robotics, autonomous ecosystems, autonomous factories, autonomous medical facilities, Boston Dynamics, DGX platform, Digital Twin, edge computing, foundation models, Foxconn, generative AI, Healthcare, humanoid robots, imitation learning, industrial systems, Isaac Lab, Isaac Sim, Jetson Thor, logistics, Manufacturing, Mega, NeMo, NVIDIA, Omniverse, OVX servers, Physical AI, productivity, Project GR00T, reinforcement learning, RGo Robotics, robot brain, robotics development, robots, simulation, smart cities, sustainability, synthetic data, training

SANTA CLARA — NVIDIA is advancing the field of physical AI by introducing specialized computing platforms designed to bring intelligence to real-world industrial systems. While generative AI has already transformed digital sectors, physical AI—artificial intelligence embodied in physical systems like robots and automated factories—has been slower to take off. However, with NVIDIA’s new suite of…
Continue Reading
modelsMeta-Llama-3-8B: Pioneering Safe and Effective AI Dialogue Solutions

AI models, Americas, Artificial Intelligence, Language model training, LLMs (Large Language Models), Open Source LLMs, Reinforcement learning, Responsible AI

July 30, 2024

RadicalShift.AI

AI Alliance, Code Shield, ethical standards, large language models, Llama 3 family, Llama Guard 2, LLMs, Meta, Meta AI, Meta-Llama-3-8B, MLCommons, optimized transformer architecture, reinforcement learning with human feedback, responsible AI development, rigorous safety protocols, RLHF, safety, supervised fine-tuning

Meta’s latest release, the Meta-Llama-3-8B, officially released on April 18, 2024, stands out as a groundbreaking model in the realm of large language models (LLMs). Designed with a keen focus on dialogue applications, this model exemplifies the company’s commitment to enhancing AI interactions through its optimized transformer architecture and rigorous safety protocols. Performance and Capabilities:…
Continue Reading
newsSanctuary AI Unveils Cutting-Edge Generation 7 of ‘Phoenix’ Robots

Americas, Artificial Intelligence, Automotive, Industrial, Machine intelligence, Reinforcement learning, Robotics

April 29, 2024

Oleg Lazarov

AGI, AI model, AI-driven robotics, artificial general intelligence, general purpose AI robotics, general purpose robotics, Geordie Rose, Phoenix Robots, Sanctuary AI, technological advancement

VANCOUVER — Sanctuary AI, a trailblazer in the field of general purpose robotics, has introduced the highly anticipated seventh generation of its flagship general purpose robot, Phoenix™. This groundbreaking release signifies a significant leap forward in advancing the capabilities of AI-driven robotics, with unparalleled improvements in both hardware and software. The latest iteration of Phoenix…
Continue Reading
newsMeta Launches Smarter, Faster, and Responsible AI Assistant Powered by the Advanced Llama 3 Model

AI assistants, Algorithms, Americas, Analytics, Artificial Intelligence, Generative AI, Internet & Online, Machine learning, Open Source LLMs, Reinforcement learning, Responsible AI

April 18, 2024

Oleg Lazarov

AI assistant, artificial intelligence, benchmarks, coding, developer tools, fine-tuning, generative AI, human feedback, large language model, Llama 3, machine learning, Meta, Meta AI, Meta ecosystem, Meta Llama 3, open community, open-source, reasoning, reinforcement learning, responsible AI, supervised fine-tuning, training data

MENLO PARK — Meta has adopted a robust and forward-thinking approach to ensure that its Meta AI and Meta Llama 3 models are developed, deployed, and used responsibly. The company recognizes the transformative power of artificial intelligence but also understands the importance of mitigating risks associated with this technology. To this end, Meta has implemented…
Continue Reading
newsGoogle Launches Gemma: Open Models for Responsible AI Development

AI assistants, Americas, Cloud, Generative AI, Internet & Online, Open Source LLMs, Reinforcement learning, Responsible AI

February 21, 2024

Sheryl Rivera

adversarial testing, AI research, AI solutions, Colab, DeepMind, developers, Gemini models, Gemma, GKE, Google, Google Cloud, Google Cloud credits, Google Cloud TPUs, Google Kubernetes Engine, Hugging Face, Hugging Face Transformers, instruction-tuned models, JAX, Kaggle, Keras 3.0, MaxText, MLOps, Model Card, NVIDIA, NVIDIA GPUs, NVIDIA NeMo™, open AI models, pre-trained models, PyTorch, red-teaming, reinforcement learning from human feedback (RLHF), researchers, responsible AI development, Responsible Generative AI Toolkit, retrieval-augmented generation (RAG), summarization, TensorFlow, TensorRT-LLM, Vertex AI

MOUNTAIN VIEW– Google is introducing a new generation of open AI models designed to support responsible AI development and assist developers and researchers globally. Gemma, built on the same research and technology behind Google’s Gemini models, aims to help foster safe, innovative AI creation. Developed by Google DeepMind and other teams across Google, Gemma is…
Continue Reading
newsNeurIPS 2020 Competitions to Leverage Procgen Benchmark and MineRL Announced

Americas, Artificial Intelligence, Entertainment, Generative AI, Generative pre-trained transformers (GPT), Machine learning, Reinforcement learning

June 21, 2020

Sheryl Rivera

AIcrowd, AlphaGo, AlphaStar, Carnegie Mellon University, DeepMind, Minecraft, MineRL, MineRL Competition, MineRL-v0 dataset, NeurIPS 2020, OpenAI, OpenAI Five, Procgen Benchmark, Procgen Competition, reinforcement learning

SAN FRANCISCO — In an exciting collaboration, OpenAI, AIcrowd, Carnegie Mellon University, and DeepMind are set to co-organize two prestigious competitions at NeurIPS 2020. These competitions, utilizing the Procgen Benchmark and MineRL, aim to drive advancements in reinforcement learning. Procgen Competition The Procgen Competition, a cornerstone of the event, challenges participants to enhance sample efficiency…
Continue Reading
newsOpenAI Introduces OpenAI LP: Pioneering a New Model for AI Investment and Impact

Americas, Artificial Intelligence, Cloud, Gaming, Generative AI, Generative pre-trained transformers (GPT), Industrial, Machine learning, Reinforcement learning, Robotics, Supercomputer

March 12, 2019

Oleg Lazarov

Adam D’Angelo, AGI, AI, Elon Musk, Greg Brockman, Holden Karnofsky, Ilya Sutskever, Khosla Ventures, OpenAI, OpenAI LP, OpenAI Nonprofit, Reid Hoffman, Sam Altman, Shivon Zilis, Tasha McCauley

SAN FRANCISCO — In a bold move to accelerate progress in artificial intelligence (AI) research, OpenAI announces the establishment of OpenAI LP, a groundbreaking “capped-profit” company designed to amplify investments in computational power and talent while staying true to its mission of ensuring the benefits of artificial general intelligence (AGI) are accessible to all. With…
Continue Reading
newsExploring Deep RL: OpenAI’s Workshop Unveils Insights and Innovations

Algorithms, Americas, Artificial Intelligence, Education, Generative AI, Generative pre-trained transformers (GPT), Industrial, Machine learning, Reinforcement learning, Robotics

February 27, 2019

Sheryl Rivera

AI, algorithm design, Blake Tucker, Daniel Ziegler, Dario Amodei, Ian Atha, Joshua Achiam, Karl Cobbe, Loren Kwan, Maddie Hall, Matthias Plappert, OpenAI, reinforcement learning, RL, RL algorithms, robotics, Ruby Chen, Spinning Up Workshop

SAN FRANCISCO — A detailed account of OpenAI‘s inaugural Spinning Up Workshop on February 2 showcases the organization’s commitment to advancing education in artificial intelligence. Drawing nearly 90 participants onsite and engaging nearly 300 more via livestream, the event attracted a diverse cohort representing various fields, including academia, software engineering, data science, ML engineering, medicine,…
Continue Reading
newsFacebook Enhances AI Research with New Montreal Lab Led by AI Expert Joelle Pineau

Algorithms, Americas, Analytics, Artificial Intelligence, Internet & Online, Investment, Machine learning, Open Source LLMs, Reinforcement learning

September 16, 2017

Oleg Lazarov

AI Ecosystem, AI Lab, AI research collaborations, artificial intelligence, Canadian research community, CIFAR, dialog systems, Facebook AI expansion, Facebook AI Research, FAIR, government policies, Joelle Pineau, McGill University, MILA, Montreal, Montreal Institute for Learning Algorithms, open source software, Reasoning and Learning Lab, reinforcement learning, startups, Université de Montréal, Yann LeCun

MENLO PARK — Facebook has announced the expansion of its AI research initiatives by opening a new artificial intelligence research lab in Montreal. This move strengthens Facebook AI Research (FAIR) by adding to its existing teams in Menlo Park, New York, and Paris. The Montreal lab will focus on cutting-edge AI projects, with a special…
Continue Reading
newsRadicalShift.AI launches

Accelerated computing, Agriculture, AI accelerators, AI models, Algorithms, Americas, Analytics, Apparel, Fashion & Beauty, Art & Culture, Artificial Intelligence, Asia, Automotive, Autonomous driving, Aviation & Aerospace, Banks, Chatbots, Chemicals, Cloud, Computer vision, Construction, Deep learning, Defence, Economy, Education, Electronics & Machinery, Energy, Gas & Oil, Entertainment, Environment, Europe, Financial, Food & Beverage, Gaming, Generative AI, Generative pre-trained transformers (GPT), Government, Graph neural network (GNN), Healthcare, Industrial, Infrastructure & Utilities, Insurance, International Trade, Internet & Online, Investment, Language model training, LLMs (Large Language Models), Machine intelligence, Machine learning, Management, Market, Marketing, Media, Metals and Minerals, Natural language processing (NLP), Neural networks, Non Profit, Open Source LLMs, Pharma & Biotech, Politics, Predictive analytics, Quantum Computing, Real Estate, Region, Reinforcement learning, Responsible AI, Retail, Retrieval-augmented generation (RAG), Robotics, Science, Security & Safety, Speech recognition and generation, Sports, Startups, Technology, Telecom, Text-to-image, Text-to-video, Transportation & Logistics, Travel

April 23, 2017

RadicalShift.AI

AI, AI industry, AI industry graph, AI industry repository, Algorithms, artificial intelligence, deployments, elevator pitch, entities, insights to knowledge, investors, jobs, LLMs, news, NLP, paradigm shift, people, products, RadicalShift.ai, RAG, scalable open platform, tools, views

RadicalShift.ai represents the paradigm shift the artificial intelligence (AI) brings upon all of us, from the way we live and work to the way we do business. To help cope with these fundamental changes across life, industries and the world in general, we are obsessively observing (30+ markets across multiple continents) and covering the AI industry while building a scalable…
Continue Reading

What is RadicalShift AI?

RadicalShift.ai represents the paradigm shift the artificial intelligence (AI) brings upon all of us, from the way we live and work to the way we do business. To help cope with these fundamental changes across life, industries and the world in general, we are obsessively observing (30+ markets across multiple continents) and covering the AI industry while building a scalable open platform aimed at people, businesses and industry stakeholders to contribute across (benefit from) the entire spectrum of the AI industry from news, views, insights to knowledge, deployments, entities, people, products, tools, jobs, investors, pitch decks, and beyond, helping build what would potentially be a resourceful, insightful, knowledgeable and analytical source for AI related news, information and resources, ultimately becoming the AI industry graph/repository.

AI.RadicalShift

Category: Reinforcement learning

researchRevealing the Mind of a Language Model: Pythia and the Transparent Science of LLMs

newsBoston Dynamics Teams Up with Robotics & AI Institute to Advance Reinforcement Learning for Atlas Humanoid Robot

newsNVIDIA Accelerates Physical AI with Advanced Platforms for Humanoid Robots and Autonomous Facilities

modelsMeta-Llama-3-8B: Pioneering Safe and Effective AI Dialogue Solutions

newsSanctuary AI Unveils Cutting-Edge Generation 7 of ‘Phoenix’ Robots

newsMeta Launches Smarter, Faster, and Responsible AI Assistant Powered by the Advanced Llama 3 Model

newsGoogle Launches Gemma: Open Models for Responsible AI Development

newsNeurIPS 2020 Competitions to Leverage Procgen Benchmark and MineRL Announced

newsOpenAI Introduces OpenAI LP: Pioneering a New Model for AI Investment and Impact

newsExploring Deep RL: OpenAI’s Workshop Unveils Insights and Innovations

newsFacebook Enhances AI Research with New Montreal Lab Led by AI Expert Joelle Pineau

newsRadicalShift.AI launches

What is RadicalShift AI?

Latest Entries

🏭 INDUSTRIES / MARKETS:

🏷️ TAGS

📂 ARCHIVE