Hardware & VRAM Requirement for Llama 4 Scout in BF16
#66
by
YongchengYAO
- opened
I have access to Nvidia 80G A100 and H100 GPUs.
- Is it possible to run inference for the
meta-llama/Llama-4-Scout-17B-16E-Instructmodel in BF16? - What is the VRAM requirement?
Found the instruction: https://github.com/meta-llama/llama-cookbook/blob/main/getting-started/build_with_llama_4.ipynb
"You'll need at least 4 GPUs with >= 80GB each"
YongchengYAO
changed discussion status to
closed