github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9769] for Gemma 3,which one has better performance, 27b(4bit) or 12b(fp16) #6387

New Issue
Closed
opened 2026-04-12 17:53:55 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 17:53:55 -05:00
Owner
Copy Link

Originally created by @zhaojigang on GitHub (Mar 14, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9769

for Gemma 3,which one has better performance, 27b(4bit) or 12b(fp16)

Originally created by @zhaojigang on GitHub (Mar 14, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9769 for Gemma 3,which one has better performance, 27b(4bit) or 12b(fp16)
GiteaMirror added the question label 2026-04-12 17:53:55 -05:00
GiteaMirror commented 2026-04-12 17:53:57 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Mar 14, 2025):

You know, this is a great question. Back of the napkin math is the 12b fp16 model is going to be larger (24GB vs 17GB), but we don't have definitive numbers on which speed/perplexity. In both cases the vision encoder/projector is going to be unquantized so will be identical between the two; only the text part is different.

<!-- gh-comment-id:2725967645 --> @pdevine commented on GitHub (Mar 14, 2025): You know, this is a great question. Back of the napkin math is the 12b fp16 model is going to be larger (24GB vs 17GB), but we don't have definitive numbers on which speed/perplexity. In both cases the vision encoder/projector is going to be unquantized so will be identical between the two; only the text part is different.
GiteaMirror commented 2026-04-12 17:53:57 -05:00
Author
Owner
Copy Link

@zhaojigang commented on GitHub (Mar 15, 2025):

You know, this is a great question. Back of the napkin math is the 12b fp16 model is going to be larger (24GB vs 17GB), but we don't have definitive numbers on which speed/perplexity. In both cases the vision encoder/projector is going to be unquantized so will be identical between the two; only the text part is different.

Okay, got it. Thank you for your reply

<!-- gh-comment-id:2726174748 --> @zhaojigang commented on GitHub (Mar 15, 2025): > You know, this is a great question. Back of the napkin math is the 12b fp16 model is going to be larger (24GB vs 17GB), but we don't have definitive numbers on which speed/perplexity. In both cases the vision encoder/projector is going to be unquantized so will be identical between the two; only the text part is different. Okay, got it. Thank you for your reply
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label question
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#6387
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?