[GH-ISSUE #8758] Deepseek v3 Activated params problem(also v2) #5684

Closed
opened 2026-04-12 16:58:42 -05:00 by GiteaMirror · 1 comment
Owner

Originally created by @looijijohn on GitHub (Feb 1, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8758

What is the issue?

i see in deepseek v2 (v2 and v3 has both active parameters model) ollama load all GGUF mode to vram memory

do we load just activated params to vram or all parametes to vram

offcource we should load all parameters to system vram

how deepseek model it in real application

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @looijijohn on GitHub (Feb 1, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8758 ### What is the issue? i see in deepseek v2 (v2 and v3 has both active parameters model) ollama load all GGUF mode to vram memory do we load just activated params to vram or all parametes to vram offcource we should load all parameters to system vram how deepseek model it in real application ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the bug label 2026-04-12 16:58:42 -05:00
Author
Owner

@rick-github commented on GitHub (Jan 5, 2026):

Ollama will load as many parameters into VRAM as it can. Anything that doesn't fit in VRAM will be loaded in system RAM.

<!-- gh-comment-id:3709773157 --> @rick-github commented on GitHub (Jan 5, 2026): Ollama will load as many parameters into VRAM as it can. Anything that doesn't fit in VRAM will be loaded in system RAM.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#5684