Source
Mirendil taps AI Hypercomputer TPUs and GPUs for pre- and post-training applications
AI lab Mirendil is utilizing Google Cloud's AI Hypercomputer infrastructure for model pre-training and post-training applications. The system combines Google TPUs and NVIDIA GPUs to accelerate AI development.
Why it matters
This allows more scientists and engineers to conduct frontier AI research faster and at a larger scale by accelerating the experimental research loop.
The details
- Mirendil is already live with a cluster of TPU v5P chips.
- NVIDIA AI accelerated computing systems will be coming online soon.
- The setup uses managed training clusters in Gemini Enterprise Agent Platform.
Show entities and relationshipsHide entities and relationships
In this article
Key connections
Google develops and manufactures TPU v5P AI accelerator chips.
Mirendil uses AI Hypercomputer
Mirendil utilizes Google Cloud's AI Hypercomputer for model pre-training and post-training applications.
Mirendil uses Google Cloud
Mirendil expands its AI systems infrastructure on Google Cloud.
Mirendil is live with a cluster of Google TPU v5P chips for model training.
Mirendil uses Gemini Enterprise Agent Platform
Mirendil utilizes managed training clusters running in Gemini Enterprise Agent Platform.
Mirendil is a partner of Google
Mirendil partnered closely with Google on end-to-end design and deployment of AI infrastructure.
Show 4 more connectionsShow fewer connections
Mirendil uses full-stack NVIDIA AI infrastructure running on Google Cloud.
Google offers NVIDIA full-stack AI accelerated computing platforms on Google Cloud.
Behnam Neyshabur is CEO of Mirendil
Behnam Neyshabur is the cofounder and CEO of Mirendil.
Mirendil uses Reinforcement Learning
Mirendil powers reinforcement learning on a massive scale for its AI research loop.
Related events
Mirendil Selects Google Cloud AI Hypercomputer
Get the weekly recap
The stories like this one, picked and explained — once a week, straight to your inbox.