-
|
From the Mar 31 NemoClaw Livestream — Multi‑agent, scaling, and long‑running claws |
Beta Was this translation helpful? Give feedback.
Answered by
zNeill
Apr 2, 2026
Replies: 1 comment
-
|
Yes, as long as you have enough GPU memory and RAM. Each claw is effectively another agent + model runtime process. On a 128‑GB Spark, two medium‑sized Nemotron instances are workable; beyond that you’ll trade off context length and KV‑cache size. For heavier multi‑agent workloads, consider multiple Sparks or off‑loading some agents to other GPUs. |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
zNeill
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Yes, as long as you have enough GPU memory and RAM. Each claw is effectively another agent + model runtime process. On a 128‑GB Spark, two medium‑sized Nemotron instances are workable; beyond that you’ll trade off context length and KV‑cache size. For heavier multi‑agent workloads, consider multiple Sparks or off‑loading some agents to other GPUs.