Lambda

A.G.I. is here! (A) lot of (G)PUs (I)nterconnected https://lnkd.in/eqdZGbmA

Local AI cloud services coming to enterprises, startups, and research labs in South Korea! 🌏 Lambda’s AI Cloud platform expands thanks to a partnership with the region’s largest telecommunications company, SK Telecom https://bwnews.pr/4dLsqXG

🔥 Meet Hermes 3: The First Full-Parameter Fine-Tuning of Llama 3.1! 🧙♂️ ✍️ We’re excited to introduce Hermes 3, the first full-parameter fine-tune of Llama 3.1 405B, brought to you by Nous Research and trained on a Lambda 1-Click-Cluster. From strategic decision-making to complex creative writing, role-playing, and Agent building—Hermes 3 excels across the board. Try your prompts immediately on https://lambda.chat/ or use our Chat Completions API - for FREE.

Look who made the list at #84 👀 We're honored to be recognized as one of the Top 100 private cloud computing companies in the world for our commitment to serving the AI/ML community: https://lnkd.in/etRHGNfs

Hear from the experts behind Lambda's AI Developer Cloud and Gretel's synthetic data platform about how their combined stack accelerates GenAI experimentation. Join us to learn how to: 🖊️ Design and iterate on a task-specific dataset using Gretel Navigator 💻 Fine-tune a Small Language Model (SLM) with Lambda Cloud on several versions of the synthetic dataset from Gretel ⏩ Speed-up innovation and reduce AI development costs Register for free: https://lnkd.in/eF4wnj8s

Good: a complete deployment guide for Llama 3.1 8B and 70B. Better: solid 8x NVIDIA H100 SXM GPU instances available to test the guide. Best: a new $2.99 / GPU / hour price for these instances. Only settle for the best: https://lnkd.in/exRb4e5j

It’s the largest open-source model ever, rivaling leading foundation models like GPT-4, GPT-4o, and Claude 3.5 Sonnet: AI at Meta's Llama 3.1 405B opens a realm of possibilities for research and commercial applications alike. At this size, a multi-node setup is a must-have to run the model at full precision (FP16). There’s a tutorial for that! https://bit.ly/3yl3mYC

Hear from teams behind the AI developer cloud Lambda and the synthetic data platform Gretel about how their combined genAI stack drives faster AI experimentation and innovation. Join us to learn how to: 🖊 Design and iterate on synthetic data using Gretel Navigator 💻 Quickly spin up world-class GPU compute via Lambda  ⏩ Speed-up innovation and reduce AI development costs When: August 13, 2024 | 10:00 am PT / 1:00 pm ET Register for free 👇

I'm excited to share a new collaborative dataset research project -- Self-Directed Synthetic Dialogues and Revisions. We set out with the goal of "creating the first at-scale constitutional AI dataset" with open language models. While doing it, HuggingFace released the first dataset, synthetic dataset has gotten far more popular, and our end result changed a bit. The dataset is a bit more creative. We (AI2, EleutherAI, SynthLabs, and compute from Lambda) are releasing a dataset with two components: 1. ~300k Dialogues (and plans for dialogues) of language models talking to themselves. 2. ~60k Critiques and revisions of when these dialogues violate certain constitutional AI principles. These are entirely new prompts and instructions, rather than applying CAI ontop of existing datasets. We use the models DBRX-Instruct, Nous Hermes Llama 2 70B (yes we'll have to redo the data), and Mistral Large v1. There are lot of positives to take away from this project (only made easier with Llama 3.1): 1. Complex synthetic data works with open language models. 2. Synthetic data is not always that sensitive. We were battling many bugs and this dataset still is decent. 3. Multi-turn data will start coming to open models. Dataset: https://lnkd.in/eVNDUNKA Report: https://lnkd.in/eWn87pKd

Lambda CFO Peter Seibold discusses how Lambda is working to provide the computational power that is driving innovation in the AI marketplace on the latest #TakingStock with Trinity Chavez. #TSTC