NewNews
October 2026 Security Incident
October 2026 Security Incident On October 15, 11:00 PM UTC we identified a security incident on the Lightning Platform involving unauthorized access to credentials.
Lightning Blog
News
NewBuild
September Recap: What We Shipped This Month
September was full speed ahead on what we shipped. If you missed our weekly recaps on LinkedIn and X, here's what happened.
NewCase Studies
How Pangram Saved 3 Engineer-Years Building AI Detectors with Lightning AI
The AI-detection startup scaled from 3 people to 25 without ever hiring a dedicated infrastructure team.
Changelog
NewChangelog
Model API deprecation: DeepSeek V4 Pro
The DeepSeek V4 Pro serverless Model APIs is scheduled to be retired at 5pm PT on September 23, at which time this model will be disabled and any requests made to it will return HTTP error code 400.
NewBuild
PyTorch Lightning checkpoints write up to 95% faster on Google Cloud Storage Rapid Buckets
We worked with Google Cloud to fix how PyTorch Lightning moves data between GPUs and object storage.
Changelog
NewChangelog
Model API deprecation: GPT-OSS 20B, 120B
The GPT-OSS 20B and GPT-OSS 120B serverless Model APIs are scheduled to be retired at 5pm PT on September 15, at which time these models will be disabled and any requests made to them will return
NewShip
August Recap: What We Shipped This Month
Our team covered a lot of ground between what we shipped and where we were in August. If you missed our weekly recaps on LinkedIn and X, here’s what happened.
Changelog
NewChangelog
Model API deprecation: Nemotron 3 Nano Omni
The Nemotron 3 Nano serverless Model API is scheduled to be retired at 5pm PT on August 19, at which time this model will be disabled, and any requests made to it will return HTTP error code 400.
NewShip
July Recap: What we shipped this month
We covered a lot of ground with what we shipped in July 2026. If you missed our weekly recaps on LinkedIn and X, here’s what happened.
NewBuild
One Team, Full Stack: Fully Virtualized H100s, Zero Compromise
How we built a performant AI-native cloud on owned H100 and B200 GPUs, with bare-metal NCCL speed inside fully managed VMs.
NewCase Studies
How Lightning AI Built NVIDIA SuperPODs for Cursor’s RL Training
When Cursor needed a partner to design, deploy, and operate the compute foundation for their next venture, they trusted Lightning AI to architect a customized NVIDIA HGX B200 SuperPOD that didn’t
NewReleases
Introducing Lightning Cloud: instant H100 access on Lightning’s fully-integrated training and experimentation cloud
Lightning AI now owns and operates its H100 fleet end-to-end. Guaranteed on-demand and spot capacity with a 1 year commitment, built for enterprise AI teams.
Changelog
NewChangelog
Model API deprecation: MiniMax M2.5, Nemotron 3 Ultra
5 and Nemotron 3 Ultra serverless Model APIs are scheduled to be retired at 5pm PT on July 13, at which time these models will be disabled and any requests made to them will return HTTP error code
NewBuild
From "Waiting for GPUs" to a Production App in a Day
Therna Biosciences traded infrastructure bottlenecks for research velocity on Lightning AI
NewNews
Why Top-Performing Teams Resort to Shadow AI
Shadow AI is when researchers build outside your approved AI infrastructure. Learn what causes it, why it is a growing enterprise risk, and how building on Lightning AI reduces your threat exposure.
Changelog
NewChangelog
Model API deprecation: Kimi K2.5, GLM 5
5 and GLM 5 serverless Model APIs are scheduled to be retired at 5pm PT on June 11, at which time these models will be disabled and any requests made to them will return HTTP error code 400.
NewReleases
Frontier reasoning at scale: NVIDIA Nemotron 3 Ultra for long-running agents on Lightning
Deploy Nemotron 3 Ultra, an open frontier reasoning model for long-running agents. Scale orchestration and reasoning with Lightning AI. No proprietary model lock-in.
Changelog
NewChangelog
Model API deprecation: DeepSeek V3.1, Llama 3.3 70B
3 70B serverless Model APIs are scheduled to be retired at 5pm PT on May 15, at which time these models will be disabled and any requests made to them will return HTTP error code 400.
NewNews
How the PyTorch Lightning Community Discovered a Supply Chain Attack and Fixed it in 42 Minutes
3. The attack targeted the distribution layer, not the source code repository.
NewBuild
Building Multimodal Agent Systems with NVIDIA Nemotron 3 Nano Omni on Lightning
Build multimodal agent systems that can see, hear, and reason using NVIDIA Nemotron 3 Nano Omni on Lightning.
NewBuild
DeepSeek V4 Alters Everything We Knew About Price-Performance Math
DeepSeek dropped two open-weight models engineers have been waiting on for months.
NewChangelog
DeepSeek V4 Pro Day 0 Support
DeepSeek V4 Pro is now available on Lightning AI. Start building with it today using a single API call.
NewBuild
What Is an Agent Harness? What Engineers Need to Understand
The term "agent harness" is being thrown around on blogs, X, podcasts and in heated debates between household names.
NewBuild
ComfyUI in the Cloud for Agencies and Creative Teams
Looking to run ComfyUI in the cloud without installing drivers or managing GPUs? Lightning lets you launch ComfyUI online in one click, with up to 80 free GPU hours and no subscription required.
NewBuild
What Is Autoresearch? The Open Source Project Running ML Experiments as you Sleep
Andrej Karpathy's autoresearch runs AI-driven ML experiments autonomously — no supervision needed. Learn how it works and run it on Lightning Studios.
NewBuild
5 Things to Do With MicroGPT to Get the Most Value Out of It
5 real ways to get value from Andrej Karpathy's microGPT, from experiment tracking to transformer intuition.
NewBuild
Run Autonomous, Self-Evolving Agents Safely with Lightning + NVIDIA OpenShell
Lightning integrates with NVIDIA OpenShell by providing high-performance frontier open model endpoints, nabling agents to leverage our compute and models safely, efficiently, and at scale.
NewBuild
What is ZeroClaw? A Rust-Powered OpenClaw Alternative
What is ZeroClaw? Discover how this OpenClaw alternative replaces heavyweight AI frameworks, and leverages Rust for secure, local-first agent workflows.
NewBuild
Working with the Lightning CLI to Manage Lightning Studios
How to manage Lightning Studios, jobs, and GPU switching entirely from your terminal using the Lightning CLI.
NewBuild
Running OpenClaw safely - Mac Mini vs Cloud Laptop.
OpenClaw is the first easy-to-use personal AI agent, but you want to run it safely. Here we compare running on a mac mini vs on the cloud.
NewNews
Lightning on Cisco AI Pods: Powering Enterprise AI End-to-End
Run enterprise AI end-to-end on Cisco AI Pods with Lightning. Train, deploy, and scale AI workloads across on-prem and hybrid cloud.
NewNews
What Is OpenClaw (Formerly Clawdbot and Moltbot)?
OpenClaw (formerly Clawdbot and Moltbot) is a viral autonomous AI agent that hit 100K stars and triggered Mac Mini sellouts.
NewNews
Why we merged Lightning AI and Voltage Park to build an AI-native cloud
Lightning AI and Voltage Park merge to build a vertically integrated, AI-native cloud - combining AI software and owned GPU infrastructure in one platform.
NewBuild
Migrating from Neptune to Lightning AI: Experiment Tracking with LitLogger
Introduction to LitLogger With Neptune shutting down, many ML teams are facing the same question: "Where do we move our experiment tracking without rewriting our code or losing history?"
NewOpinion
Fault-tolerant AI inference despite Cloudflare outage
Cloudflare is down again, and I just want to rant a bit about how to think about protecting against these scenarios if you are new to AI inference.
NewCase Studies
Infineon Builds a Language Model Small Enough to Run on a Microcontroller in 4 Months
Infineon compressed 1.5-2 years of development into 4 months with Lightning AI.
NewNews
Lightning AI Launches First GPU Marketplace Uniting NeoClouds and Hyperscalers, Cutting AI Costs by 70%
For the first time, teams can access high-performance GPUs with a full-featured platform. No tradeoffs between flexibility, speed, or scale.
NewBuild
Build faster with Lightning: a developer-first GPU marketplace
Access high performance GPUs with a full-featured platform. No tradeoffs.
NewCase Studies
Slalom Accelerates Model Development and Cuts Setup Time Using Lightning AI
Slalom sped up model development and reduced setup overhead, empowering a small team to deliver big results.
NewOpinion
GPT 5 vs GPT OSS 20B vs GPT OSS 120B
I put OpenAI’s newest models, GPT-5, GPT OSS 120B, and GPT OSS 20B, head-to-head on a real production fraud detection task. The goal was to find the fastest model that doesn’t sacrifice accuracy..
NewCase Studies
How this Team Won the LeRobot Hackathon by Folding Shirts and Reimagining Robotics
A three-person team taught a robot arm to fold t-shirts and sort medicine, finishing 1st in the U.S. and 2nd globally, all built on Lightning AI.
NewInference
From Flask to vLLM: How Model Inference has evolved (2017-2025)
. from Flask chatbots to full AI systems. In this post, I share what’s changed in the model inference world, what’s worked for me, and how I think about serving models in 2025.
NewCase Studies
How NNAISENSE Built an AI-Powered Hedge-Fund And Cut Infrastructure Costs By 35%
A hedge fund startup simplified infrastructure and focused on strategy, saving time and money with Lightning AI.
NewCase Studies
How Columbia Labs Accelerates Neuroscience Research and Democratizes AI with Lightning AI
Dr. Matt Whiteway’s team at Columbia University designed and shared their pose estimation tool with labs everywhere, without installation, GPU setup, or onboarding headaches.
NewCase Studies
Sentinal4D Advances Cancer Research with Lightning AI, Achieving 2x Faster Development
Sentinal4D halved their oncology drug discovery time by optimizing training, simplifying compute access, and improving global team collaboration with Lightning AI.
NewCase Studies
Reticular Sped Up Drug Discovery Research From 6 to 3 Months with Lightning AI
Reticular, a biotech startup, accelerated drug discovery research by simplifying collaboration and speeding up model experimentation
NewCase Studies
Inworld AI Cut Months Off Voice AI Development with Lightning AI
Inworld AI accelerated Voice AI while reducing costs and advancing time-to-market.
NewCase Studies
Synthera AI Brings Generative Simulations to Finance 5 Months Faster with Lightning AI
An AI startup skipped 5 months of infrastructure work to focus on building what mattered.
NewTraining
B200 vs H100 for model training
Most teams either overpay for B200s or underuse them. We benchmarked B200 vs H100 to give you a clear, no-fluff guide to making the right call for your next training run.
NewCase Studies
How GenreX brought AI music to market 70% faster
GenreX is using AI to reinvent how music is made for games, ads, and creators everywhere.
NewCase Studies
Decoding the brain by training 1000 AI models to predict movement
At the Paninski lab at Columbia University, researchers trained 1,000 AI models in just 2 weeks to accelerate research that connects animal movement to brain activity and brings us closer to
NewCase Studies
Reducing carbon emissions faster at the MIT Sustainable Design Lab
MIT’s sustainability lab helps cities plan and reduce their carbon footprint with AI.
NewNews
Lightning AI Joins AI Alliance To Advance Open, Safe, Responsible AI
Company Joins International Community of AI Organizations, Led by IBM and Meta, That Includes Leading Startups, Academia, Research and Government Organizations
NewInference
8-bit Quantization with Lightning Fabric
Learn how to shrink massive models like Llama 2 and SDXL with a single line of code using Lightning Fabric’s 8-bit quantization plugin.
NewBuild
Quickstart to Lightning Fabric
Scaling raw PyTorch is painful: too much boilerplate, too many edge cases. PyTorch Lightning can feel too opinionated when you want full control.
NewInference
4-Bit Quantization with Lightning Fabric
Readers will learn the basics of Lightning Fabric’s plugin for 4-bit quantization.
NewTraining
Doubling Neural Network Finetuning Efficiency with 16-bit Precision Techniques
This guide targets PyTorch model training, illustrating how you can adjust the floating point precision to drastically enhance training speed and halve memory consumption, all without compromising
NewBuild
Step-By-Step Walk-Through of Pytorch Lightning
Training deep learning models in raw PyTorch gets repetitive fast. This walkthrough trains a CIFAR-10 image classifier with PyTorch Lightning, plus loggers and callbacks.
NewBuild
PyTorch Lightning for Dummies – A Tutorial and Overview
You’ll learn to use PyTorch Lightning’s Core API features by completing an applied project to train a Language Transformer written in PyTorch on the WikiText2 dataset.
NewBuild
From PyTorch to PyTorch Lighting: Getting Started Guide
Managing training loops, device placement, logging, and scaling in raw PyTorch quickly becomes tedious and error-prone as projects grow.
NewTraining
Finetuning LLMs with LoRA and QLoRA: Insights from Hundreds of Experiments
LoRA is one of the most widely used, parameter-efficient finetuning techniques for training custom LLMs.
NewTraining
Scaling Large (Language) Models with PyTorch Lightning
In this blog, you will learn about techniques to train large models like Llama (or any LLM) and Stable Diffusion using distributed training strategy FSDP with PyTorch Lightning.
NewInference
What is Quantization
Learn how quantization reduces the memory footprint of models like Llama 2 as much as 4x!
NewOptimize
Torchmetrics v1.2
What clustering is, why it’s a useful measure, and newly added metrics that can be used with code samples.
NewInference
Optimizing LLMs from a Dataset Perspective
Discover new research directions to improve Large Language Models (LLMs) and learn how to enhance the performance of instruction-finetuned LLMs by concentrating on higher-quality data and exploring
NewBuild
When to Use PyTorch Lightning or Lightning Fabric
Discover how to use PyTorch Lightning or Lightning Fabric to pretrain your custom model checkpoints or finetune LLMs like CodeLLaMA and LLama 2.
NewTraining
The NeurIPS 2023 LLM Efficiency Challenge Starter Guide
Large language models (LLMs) offer one of the most interesting opportunities for developing more efficient training methods.
NewNews
Stability AI announced SDXL 1.0 – Trained using PyTorch Lightning
We are excited to share that PyTorch Lightning continues to power Generative AI research for training large foundation model, SDXL from Stability AI, by simplifying distributed training for massive
NewBuild
How to build a chatbot using open-source LLMs like Llama 2 and Falcon
In this blog, we’ll learn how to build a chatbot using open-source LLMs. We will be using Lit-GPT and LangChain. Lit-GPT is an optimized collection of open-source LLMs for finetuning and inference.
NewNews
NeurIPS LLM Efficiency Challenge
Finetune Lit-GPT on 1 GPU in 1 day and get a chance to win a speaker slot at NeurIPS 2023.
NewData Science
Tabular Classification with Lightning
Learn how to gain back research time by leveraging PyTorch Lightning for over 100 inbuilt methods, hooks, and flags that save you engineering hours on heavy lifts like distributed training in
NewOptimize
TorchMetrics — a short history to 100+ metrics
As a celebration of v1.0.0 of Torchmetrics being released, in this blogpost we will discuss the history of Torchmetrics and how the interface have evolved over time.
NewOptimize
Torchmetrics v1.0: Visualize model performance with 100+ metrics
0, is now publicly available.0, we have also managed to implement over 100 metrics in torchmetrics. In this blogpost we go over some of the new features and improvements to the library.
NewOptimize
Optimizing Memory Usage for Training LLMs and Vision Transformers in PyTorch
Peak memory consumption is a common bottleneck when training deep learning models such as vision transformers and LLMs.
NewOptimize
Faster PyTorch Training by Reducing Peak Memory (combining backward pass + optimizer step)
A generic four line code change can reduce peak memory by 20-40% when training large models.
NewTraining
Efficient Initialization of Large Models
One of the challenges with LLMs is their cost and large memory footprint.1 release, we introduce new features that optimize all three stages of LLM usage: pretraining, finetuning, and inference!
NewTraining
Finetuning Falcon LLMs More Efficiently With LoRA and Adapters
Using parameter-efficient finetuning methods outlined in this article, it’s possible to finetune an open-source LLM like Falcon in 1 hour on a single GPU instead of a day on 6 GPUs.
NewNews
Falcon – A guide to finetune and inference
In this blog you will learn about the latest open-source large language model Falcon.5 GB of GPU memory!
NewOpinion
How Military Veterans Can Get into AI
Learn how to gain the technical skills to break into AI after a successful military career.
NewNews
Lang Segment Anything – Object Detection and Segmentation With Text Prompt
Our community member Luca Medeiros shows how he leveraged the Segment Anything Model from Meta AI and built his lang-segment-anything library for object detection and image segmentation based on text
NewTraining
How To Finetune GPT Like Large Language Models on a Custom Dataset
Learn how to finetune large language models (LLMs) on a custom dataset. We will be using Lit-GPT, an optimized collection of open-source LLMs for finetuning and inference.
NewTraining
How To Train Reinforcement Learning Model To Play Game Using Proximal Policy Optimization (PPO) Algorithm
Learn how to build and train a Reinforcement Learning model with PyTorch and Lightning Fabric.
NewOptimize
Accelerating Large Language Models with Mixed-Precision Techniques
Training and using large language models (LLMs) is expensive due to their large compute requirements and memory footprints.
NewNews
The Ultimate Battle of Language Models: Lit-LLaMA vs GPT3.5 vs Bloom vs …
Large Language Models (LLMs) are taking center stage, wowing everyone from tech giants to small business owners.
NewTraining
Parameter-Efficient LLM Finetuning With Low-Rank Adaptation (LoRA)
In the rapidly evolving field of AI, using large language models in an efficient and effective manner is becoming more and more important.
NewNews
How to Contribute to Lit-GPT and Lit-LLaMA
Learn about the variety of ways to contribute to Lit-LLaMA, Lightning AI’s open-source rewrite of LLaMA.
NewTraining
Understanding Parameter-Efficient Finetuning of Large Language Models: From Prefix Tuning to LLaMA-Adapters
Learn how popular parameter-efficient finetuning methods for LLM work: prefix tuning, adapters, and LLaMA-Adapter.
NewTraining
Accelerating LLaMA with Fabric: A Comprehensive Guide to Training and Fine-Tuning LLaMA
In this tutorial, we will learn how to train and fine-tune LLaMA (Large Language Model Meta AI). Lit-LLaMA, a rewrite of LLaMA, can run inference on an 8 GB consumer GPU.
NewOpinion
Why We’re Building Lit-LLaMA
With Lit-LLaMA, our goal is to provide a clean, fully open, non harmful implementation of the LLaMA source code that can be rapidly developed alongside the community.
NewNews
How to Contribute to Lightning
Contributing to open-source projects is a great way to improve your coding skills
NewTraining
Finetuning LLMs on a Single GPU Using Gradient Accumulation
Learn how to use gradient accumulation to train models with large batch sizes in order to work around hardware limitations when GPU memory is a concern.
NewOptimize
How to Speed Up PyTorch Model Training
In this tutorial, we will finetune a DistilBERT model, a distilled version of BERT that is 40% smaller at almost identical predictive performance
NewOptimize
Accelerate PyTorch Code with Fabric
Use Lightning Fabric to train and accelerate a PyTorch model using mixed precision and distributed training.
NewBuild
Introduction to Lightning Fabric
Lightning Fabric is a new, open-source library that allows you to quickly and easily scale models while maintaining full control over your training loop.
NewTraining
Training Compiled PyTorch 2.0 with PyTorch Lightning
How to use PyTorch 2.0 and train a compiled model with PyTorch Lightning 2.0. Find the full code used in this tutorial here.
NewTraining
Training Compiled PyTorch 2.0 with PyTorch Lightning
PyTorch 2.x aims to push the performance with model compilation further while maintaining its Pythonic nature and backward compatibility.
NewReleases
Introducing PyTorch Lightning 2.0 and Fabric
Today, we’re introducing PyTorch Lightning 2.0—as well as Fabric, a new library—to continue unlocking unprecedented scale, collaboration, and iteration for researchers and developers.
NewReleases
PyTorch Lightning 2.0 FAQ
Answers to common questions about PyTorch Lightning 2.0: what changed, what is simpler, and what Lightning Fabric adds for finer control.
NewOpinion
Are Consistency Models the Next Big Thing?
OpenAI is creating an entirely new class of generative models for images, and researchers are wondering how language models do in-context learning.
NewNews
Microsoft’s Kosmos-1 Claims to Solve IQ Tests and OpenAI Introduces ChatGPT and Whisper APIs
Developers can now integrate ChatGPT and Whisper models into their apps and products through OpenAI’s API. Microsoft researchers introduced Kosmos-1, a MLLM that claims to solve IQ tests.
NewOptimize
Using DeepSpeed to Optimize Models
A discussion of various techniques that you can use to get the most out of your deep learning models.
NewReleases
Be the First to Know: Lightning Open Source Launch
Be the First to Hear About What is Launching on March 15th! We’ve been hard at work implementing user feedback into the next release of PyTorch Lightning, and we’re excited to unveil it on March 15th.
NewBuild
Object Detection on the Cloud with YOLOv8
Posted on February 28, 2023 by Aniket Maurya & Ayush Chaurasia - Community , Tutorials
NewNews
Why Google’s Researchers are Intentionally Poisoning Datasets and More
Can intentionally poisoning web-scale training datasets be a practical approach to preventing risk? How can model predictions and their intended usage be better aligned?
NewCase Studies
Nixtla's AI models make better forecasts with less data
Nixtla's new pretrained AI models let you forecast the future with almost no historical data.
NewNews
AI Just Flew a Fighter Jet for 17 Hours, Commanded Super Mario Bros and More
ML twitter is in a flurry over ToolFormer, a model that is getting LLMs to use tools in a self-supervised manner.
NewNews
Google’s Bard Made a $100 Billion Mistake and Bing’s New Chat Capabilities Delights Users
The battle of the chatbots continues with Bard’s reception peaking and falling in a single week while Bing’s new interface is causing a flurry of excitement.
NewShip
Accelerate Serving Stable Diffusion by Optimizing the Inference Scheduler
Learn how to accelerate serving models with sequential inference steps, like Stable Diffusion.
NewNews
Diffusion Models Under Fire for Privacy Concerns and More
Researchers are raising a flag about the privacy risks of the latest diffusion models and the AI detection model created by Princeton student is now GA.
NewData Science
How to Evaluate Machine Learning Models
As machine learning models continue to gain widespread adoption in industries ranging from healthcare and manufacturing to retail and logistics, it has become increasingly important that we evaluate
NewNews
Open-Source Implementations of ChatGPT’s Training Algorithm Are Released and more
Researchers are hoping watermarking language models can help stop plagiarism while LAION and CarperAI are helping the public replicate ChatGPT on their own.
NewShip
Serve Stable Diffusion Three Times Faster
Serving large models in production with high concurrency the ability to serve multiple simultaneous inference requests and throughput
NewNews
Princeton Student Creates Tool To Detect ChatGPT and an AI Bot Just Beat the Rocket League Elite
Thanks to researchers from DeepMind, you can now build your own transformer models from scratch!
NewBuild
머신러닝 훈련 및 배포 파이프라인을 만드는 방법
Lightning을 이용하여 어떻게 모델 훈련 및 배포 파이프라인을 만들 수 있는지 배웁니다. 해당 파이프라인은 커스터마이징 할 수 있고, monitoring, data warehouse, feature store 등 여러 기능과 통합되어 있으며, production에 배포되어 있습니다.
NewTraining
Train Scikit-learn Models on the Cloud
The scale of machine learning datasets can often become large enough that migrating to a cloud provider for training, storing checkpoints, and logs becomes necessary.
NewNews
ChatGPT Inspires Virtual AI “Wife”, DeepMind Introduces DreamerV3, and OpenAI Teases Plans for Premium Chat
Researchers demonstrate surprising differences in causality-based localization vs. knowledge editing in language Models.
NewNews
The World’s First Robot Lawyer Is Here, Microsoft Hopes ChatGPT Will Make Bing Smarter and Scaling Model Serving in Prod Just Got Easier
Can we we compress large language models for better performance? Can we rethink LLM processing? These are just a few of the questions researchers are asking themselves in the new year.
NewOpinion
What WON’T Happen in AI in 2023
For this special end-of-year edition of our newsletter, we are poopin’ on parades and kicking over crystal balls. Here is what we’re betting won’t happen in the coming year:
NewShip
Scale Model Serving with Dynamic Batching and Autoscaling
How to scale model serving in production with dynamic batching and autoscaling.
NewBuild
How to Build a Machine Learning Training and Deployment Pipeline
Learn how you can use Lightning to build a model training and deployment pipeline that is customizable, integrated with tools (like monitoring, data warehouses, and feature stores), and deployed in
NewNews
Sebastian Raschka Launches a Deep Learning Course, NFL Wants ML to Predict Injuries, and Researchers Find 4 Sub-Phenotypes of Long COVID
Researchers in France have identified the four sub-phenotypes of long COVID while researchers in the US just released the largest federated ML effort to date.
NewShip
Deploy OpenAI Whisper as a Cloud Product
A New ML Model, a New Startup Opportunity — give Echo, our free transcription app powered by OpenAI’s Whisper, a try!
NewNews
OpenAI’s New Chatbot, Stable Diffusion 2.0, and the TLDR on the Top Papers at NeurIPS 2022
Researchers are testing the limits of language models and datasets while diffusion models continue to evolve.0 on your own servers and have a surprisingly human conversation with OpenAI’s ChatGPT.
NewNews
NVIDIA and Microsoft Team Up To Build an AI Supercomputer, Meta Releases Galactica and Sony Patents a New ML System
Robot dogs and LLMs are getting more capable. NVIDIA and Microsoft team up on an AI supercomputer, and Meta releases Galactica.
NewShip
Как деплоить диффузионные (diffusion) модели
This Russian-language blog post, completed by a member of the Lightning community, is a translation of the English original
NewNews
Colossal-AI Accelerates AIGC, Snowpark for Python Is GA and the New Beta Release of NVIDIA Omniverse
Researchers are improving audio quality with real-time target sound extraction and complex music mixing.
NewNews
Meta AI Releases a Metagenomic Atlas, Space Force Is Back With AI, and DALL·E API Now Available in Public Beta
Meta AI’s researchers are celebrating the release of an open atlas of 617 million metagenomic protein structures, AI is being deployed to forecast satellite failures in orbit, and developers can now
NewReleases
Lightning 1.8: Colossal-AI, Secrets for Apps, and more
We’re excited to announce the release of Lightning 1.8 ⚡️ (release notes).
NewNews
Bumble Is Blocking 🍆 Pics, DeepMind Proposes Algorithm Distillation and Shutterstock Pairs Up with DALL-E 2
RThis week, researchers are improving the audio quality of your streaming services and continuing to improve reinforcement learning algorithms.
NewTraining
Guide to Distributed Training
In this tutorial, you’ll learn how to scale models and data across multiple devices using distributed training.
NewNews
Steve Jobs is Resurrected, Meta Is Translating Unwritten Languages and AI is Running for Office
Researchers continue to push the limits of diffusion models, Steve Jobs returns for a conversation with Joe Rogan and Jack Dorsey’s new decentralized social network is open for beta users.
NewNews
Zuckerberg’s Avatar Gains “Social Presence”, AI Learns How to Play Nice, and Building Diffusion Models Just Got Way Easier
Lightning is handing out the blueprints of Stable Diffusion and researchers are helping AI become better at cooperating with humans in strategic game play.
NewShip
How to Deploy Diffusion Models
In this tutorial, you’ll learn how to deploy diffusion models at scale and build a text-to-image generator
NewNews
ICLR Releases Submissions for 2023, The White House Pens an “AI Bill of Rights” and Meta’s Make-A-Video Excites Creatives
This week, the announcement of several text-to image, voice and video generation models are delighting researchers and creatives alike.
NewTraining
How to Enable Native Fully Sharded Data Parallel in PyTorch
this tutorial teaches you how to overcome hardware constraints when training large models using PyTorch’s new model sharding strategy
NewData Science
Train Machine Learning Models with Synthetic Data
The Lightning App built with NVIDIA Omniverse Replicator (try it for free today!) lets you generate synthetic data to enhance the training and performance of your AI models.
NewNews
PyTorch Moves to Linux Foundation, Chinese Corpus for NLP…
Can a single policy control the movement of hundreds of different robots? Is it possible to make 3D mapping more accurate and cheaper at the same time?
NewTraining
How to Enable GPU-Accelerated Training on Apple Silicon in PyTorch
this tutorial shows you how to train models faster with Apple’s M1 or M2 chips.
NewNews
PyTorch Team at Lightning – Welcome Mike Ruberry
Today, I’m thrilled to share that Mike Ruberry, one of PyTorch’s core technical leads, will be joining Lightning.
NewReleases
Lightning 1.7: Apple Silicon, Multi-GPU and more
We’re excited to announce the release of PyTorch Lightning 1.7 ⚡️ (release notes!)
NewReleases
Lightning 1.6: Habana Accelerator, Bagua Distributed, Fault-Tolerance Improvements
6 with support for Intel’s Habana Accelerator, new efficient DDP strategy (Bagua), manual Fault-tolerance, and other stability and reliability changes.
NewData Science
Generate Synthetic Data with NVIDIA’s Omniverse Replicator Lightning App (beta)
Start using the Omniverse Replicator Lightning App for free today. At SIGGRAPH , NVIDIA and Lightning AI debuted the beta of the Omniverse Replicator Lightning App .
NewNews
Lightning Welcomes Jatheen, Priya, Mike, and JP!
Today, we’re excited to announce four senior executive hires who will enable us to advance our mission of accelerating an AI-powered world.
NewBuild
Using Redis in a Machine Learning Application
How to use Redis in a database system for machine learning applications
NewBuild
How to Build Apps on Lightning AI
At June’s inaugural Lightning Developer Conference, Lightning AI’s Staff Product Manager Eden Afek explored how you can bring your machine learning models to life through AI-powered Lightning Apps.
NewBuild
How To Build a Super Resolution GAN Demo in Lightning AI
Building and training your deep learning and machine learning models is great, but what do you do once you’ve worked out the kinks?
NewBuild
Build an Interactive Research Poster with Lightning AI
An Easy Way to Demonstrate Your Research A research poster is an attractive visual presentation of research that displays the information in a clear and concise manner.
NewBuild
Build with Lightning and Win! Introducing the Storm Chaser Challenge Series
Our first Storm Chaser challenge series runs July 20 to August 3, 2022. Complete the tasks to win a $1,000 Visa gift card.
NewBuild
Building a GAN (Generative Adversarial Network)
Train And Explore Generative Adversarial Networks (GANs) Interactively With Lightning Apps
NewOpinion
Stop Wasting Time on MLOps
The Fragmented World of Machine Learning/MLOps At June’s inaugural Lightning Developer Conference, Founder and CEO William Falcon’s keynote explored the state of artificial intelligence, and how it’s
NewBuild
Recurrent Neural Networks (RNNs)
A StatQuest Tutorial When you don’t always have the same amount of data, like when translating different sentences from one language to another, or making stock market predictions from different
NewBuild
An Introduction to Lightning AI
What Lightning AI is, how it differs from Grid.ai, and how the open-source Lightning framework fits into the platform.
NewNews
Grid.ai Becomes Lightning AI
Grid.ai is now Lightning AI. Here is what the rename means and why we are unifying the AI development lifecycle, along with our $40M funding round.
NewTraining
Performance Notes Of PyTorch Support for M1 and M2 GPUs
Apple's M1/M2 chips, known for strong performance and energy efficiency, now support GPU acceleration in PyTorch, and while their GPU RAM usage is higher than CUDA GPUs, training with adjusted batch
NewCase Studies
Scaling Accelerated Drug Discovery
SyntheticGestalt, an AI startup based in London and Tokyo, is developing an automatic system to make valuable drug discoveries en masse.
NewNews
What is Grid
The Grid platform enables users to quickly iterate through the model development life cycle by managing the provisioning of machine learning infrastructure on the cloud.
NewBuild
Introduction to PyTorch
A StatQuest Tutorial PyTorch is one of the most popular tools for making Neural Networks. This StatQuest walks you through a simple example of how to use PyTorch one step at a time.
NewCase Studies
Solving Complex Macroeconomic Problems With Machine Learning
How a team at The University of British Columbia was able to quickly train models and obtain results faster than ever before
NewNews
Lightning AI Acquires Tensorwerk
A Letter From Luca Antiga, Co-Founder and CEO of Tensorwerk Since its inception, Tensorwerk has been about people.
NewCase Studies
How Computer Vision Researchers Use Grid For Their Infrastructure Requirements
The store of Clutterbot, a New Zealand-based startup that builds autonomous cleaning robots
NewCase Studies
Avoid Building and Maintaining an MLOps Training Environment
Using sophisticated machine learning techniques, the AutoDevTech platform accelerates a team’s efficiency by providing valuable insights into their development process in the context of industry
NewCase Studies
How Neuroscientists Are Using AI To Understand Behavior
Before discovering Grid, the team managed all their cloud experiments themselves on AWS instances using boto3 scripts and vanilla PyTorch.
NewCase Studies
Managing Machine Learning With Limited Resources
In the race to implement machine learning, businesses of all sizes look for more accurate predictions that will allow them to stay ahead and differentiate themselves from the competition.
NewOpinion
Why we open sourced our documentation
Grid is a scalable machine learning platform that empowers researchers and data scientists to rapidly iterate on ideas without infrastructure hassles, while also enabling community-driven
NewTraining
Get Full Control of your Training Using Callbacks
Lightning has a technique to add arbitrary self-contained programs to your training which are called callbacks.
NewTraining
Avoid Overfitting in Machine Learning
Using the EarlyStopping callback, Lightning will automatically stop training once a specific metric has stopped improving.
NewTraining
Reload DataLoaders Every Epoch
How Lightning Helps You Reload Your Data on Every Epoch Lightning is a lightweight PyTorch wrapper for high-performance AI research that reduces the boilerplate without limiting flexibility.
NewTraining
Truncated Back-propogation Through Time
How Lightning Helps You With Long Sequences Lightning is a lightweight PyTorch wrapper for high-performance AI research that reduces the boilerplate without limiting flexibility.
NewTraining
Exploding And Vanishing Gradients in Lightning
How Lightning Identifies Exploding and Vanishing Gradients Lightning is a lightweight PyTorch wrapper for high-performance AI research that reduces the boilerplate without limiting flexibility.
NewTraining
Automatic Learning Rate Finder in Lightning
How To Let Lightning Find the Best Learning Rate Lightning is a lightweight PyTorch wrapper for high-performance AI research that reduces the boilerplate without limiting flexibility.
NewTraining
Automatic Batch Size Finder in Lightning
How To Let Lightning Automatically Figure Out Batch Size Lightning is a lightweight PyTorch wrapper for high-performance AI research that reduces the boilerplate without limiting flexibility.
NewTraining
Lightning Training Intro
How To Get Started Training With Lightning Lightning is a lightweight PyTorch wrapper for high-performance AI research, that reduces the boilerplate without limiting flexibility.
NewReleases
Lightning 2.1: Train Bigger, Better, Faster
Lightning AI is excited to announce the release of Lightning 2.1 ⚡.The theme this time around is “bigger, better, faster.”
NewBuild
Vibe Coding My Way to a Boltz-2 Structure Prediction App
In this blog post, Amir Momen-Roknabadi walks through his efforts to "vibe code" his way into creating an app to run Boltz-2 and deploy it to the Lightning AI platform.
NewBuild
Building Reliable Agent Systems with NVIDIA Nemotron 3 Super on Lightning
Learn how to build multi-agent systems with NVIDIA Nemotron 3 Super on Lightning.
No blogs yet