Nebius’ Post

Big news from the world of AI and model training! 🚀 At Tracto, we’re working to make serverless compute accessible to every AI builder. Want to see what’s possible? Pleias, a foundational AI research lab, trained a 1 billion-parameter language model on Tracto. They used our GPU-powered platform to process massive datasets and run their model training—no infrastructure headaches and parsing thru K8s config files. Read the full story here: https://lnkd.in/dqJg4uDn Drop me a note if you are curious about what we're building. Onward!

TractoAI was built by Nebius as a serverless GPU cloud platform to enable AI developers to focus on training and fine-tuning AI models. Pleias, a leading French AI research lab, partnered with TractoAI to train their 1B LLM. Intrigued? Read the full story 🔗 https://lnkd.in/eBsRZpUE

Big news from the world of AI and model training! 🚀 At Tracto, we’re working to make serverless compute accessible to every AI builder. So what? Pleias, a foundational AI research lab, trained a 1 billion-parameter language model on Tracto. They used our GPU-powered platform to process massive datasets and run their model training—no infrastructure headaches and parsing thru K8s config files. Read the full story here: https://lnkd.in/dqJg4uDn Drop us note if you are curious about what we're building. Onward!

Phind-70B: Closing Code Quality Gap with Speed and Accuracy Phind has introduced Phind-70B, a model that aims to close the code quality gap with GPT-4 Turbo while running four times faster. It’s designed to provide high-quality answers for technical topics quickly, enhancing the user experience for developers. The model is based on the CodeLlama-70B model and has been fine-tuned on an additional 50 billion tokens. It supports a context window of 32K tokens and scores 82.3% on HumanEval. Phind-70B is also faster than GPT-4 Turbo, running at 80+ tokens per second. The model is available for use without a login, and higher limits are available with a Phind Pro subscription. #Phind #Phind70B #GPT4Turbo #CodeQuality #DeveloperExperience #CodeLlama #TechnicalTopics #AI #MachineLearning #DeveloperTools #PhindPro

“AI’s new backbone is here” - HackerNoon io.net clusters GPUs across multiple locations into a DePIN and enables the provision of on-demand, low-latency compute at the scale needed by AI startups and machine learning engineers. io.net is igniting a shift towards more efficient and flexible computing for AI companies by building the world's largest decentralized compute network. https://lnkd.in/eqJNGHDw

Exa is building a tool that allows AI models to perform something like a web search, but with an AI-native twist. The cofounders bought a million dollars worth of GPUs (which were easier to get in those days) and, using a vector database and embeddings (not a classic transformer-based LLM), they began to build an machine learning model trained to natively understand links rather than words and sentences. #AI #Google #Exa #Nvidia #YCombinator #Lightspeed

Its becomming evident that we need an instance with a gpu. Normally, we would go to renouned cloud providers. After having selected the necessary parameters of the instance. Right now we are faced with tge fact that we cannot deploy this method. due to the fact that one needs to request a quota, and this process involves talking to a human manager. This would entail checking their security team, etc. Even after such a tedious and lengthy process aimed at reaching the quota, we are likely to be faced with the fact that there are no free resources. Culminating in a wait time in order to catch the moment in which the relevant resources become available. Based on the discussion above, a relevenat solution pertaining to... "Beyond Traditional Cloud Services: This new network seeks to transcend the limitations faced by traditional cloud providers, offering a more integrated and efficient solution."

At #NVIDIA’s #GTC2024, Pure Storage made several big announcements, including these two... 💯🌟 1️⃣ Optimize GenAI Apps with Retrieval-augmented Generation from Pure Storage and NVIDIA: https://lnkd.in/exuMVD6N 🏆🚀 2️⃣ Pure Data Storage Platform for AI Expands with NVIDIA OVX Server and Full-stack AI Solutions: https://lnkd.in/e6ztfm67 ————— #DataStrategy #AIStrategy #AI #CDO #CTO #RAG #GenerativeAi #PureStorage #MachineLearning #DataScience #BigData #DataManagement #DigitalTransformation

I love Google's Colab. Why? ➤ It's far easier to type in a few letters of the URL than it is to launch an IDE and do the set up for writing code. ➤➤ Which makes it faster to satisfy your curiosity on a syntax or answer "will this work?" questions. ➤➤➤ Which makes me do more little tests and learn more. Yeah on it being faster. #python #experimenting #thinkingskills

As AI and ML models become increasingly complex, efficient GPU utilization is crucial for performance and cost-effectiveness. This webinar delves into advanced techniques such as time-slicing, Multi-Instance GPU (MIG), Multi-Process Service (MPS), and other best practice strategies available for optimizing GPU usage on Kubernetes. https://lnkd.in/e8Fskqwv