Lyceum
Inference
  • Serverless inference
  • Dedicated endpoints
  • Model library
  • Inference pricing
  • API documentation
Compute
  • GPU virtual machines
  • Serverless training
  • Dedicated GPU clusters
  • Compute pricing
Pricing Docs
Company
  • About
  • Team
  • Careers
  • Trust centre
  • Magazine
Log in Start building Welcome back, Open dashboard
Inference
  • Serverless inference
  • Dedicated endpoints
  • Model library
  • Inference pricing
  • API documentation
Compute
  • GPU virtual machines
  • Serverless training
  • Dedicated GPU clusters
  • Compute pricing
Pricing Docs
Company
  • About
  • Team
  • Careers
  • Trust centre
  • Magazine
Start building Log in Open dashboard
  1. Home
  2. › Magazine
  3. › GPU Compute
  4. › Training Infrastructure
  5. › Fine-Tuning

subcluster

Fine-Tuning

2 articles

Articles

May 20, 2026

LoRA vs Full Fine-Tuning Memory Cost: VRAM Math

You have a 24GB GPU and an 8B model. The math says it should fit, but your training script crashes with an OOM error before the first epoch. We break down the exact VRAM requirements for full fine-tuning versus LoRA.

May 18, 2026

FP8 Training on H100: Benchmarks and Memory Savings

Training a 70-billion parameter model in BF16 requires hundreds of gigabytes of GPU memory. Shifting to FP8 precision on NVIDIA H100s halves the bytes per element for the tensors actually held in FP8, master weights and optimizer states stay in higher precision, and NVIDIA's NeMo measurements show 1.30x throughput on Llama 3 8B and 1.43x on Llama 3 70B versus BF16.

Other subclusters

Distributed Runs 3 Checkpointing 1
← All articles
European AI infrastructure.
Berlin and Zurich.
Live status
  • Docs
  • Models
  • Pricing
  • Trust
  • Careers
  • Contact

© 2026 Lyceum Technology Germany GmbH

  • Privacy
  • Terms
  • Imprint

Get started with GPU compute in minutes

Book a Demo

Cookies

We use cookies to measure traffic. Rejecting keeps everything working. Learn more