github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #6303] Llama 3.1 405B fix-update #3952

New Issue
Closed
opened 2026-04-12 14:49:41 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 14:49:41 -05:00
Owner
Copy Link

Originally created by @gileneusz on GitHub (Aug 10, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6303

The update reduces memory usage, maintaining the same quality of the previous version. 🎉

Screenshot 2024-08-10 at 20 04 09

Link to updated model:

https://huggingface.co/meta-llama/Meta-Llama-3.1-405B

Screenshot 2024-08-10 at 20 04 21
Originally created by @gileneusz on GitHub (Aug 10, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6303 The update reduces memory usage, maintaining the same quality of the previous version. 🎉 <img width="597" alt="Screenshot 2024-08-10 at 20 04 09" src="https://github.com/user-attachments/assets/a4c9f9c6-7598-43a7-9967-b7f32044b269"> Link to updated model: https://huggingface.co/meta-llama/Meta-Llama-3.1-405B <img width="601" alt="Screenshot 2024-08-10 at 20 04 21" src="https://github.com/user-attachments/assets/28844f00-f894-4ea0-835f-ad58e92ef22c">
GiteaMirror added the model label 2026-04-12 14:49:41 -05:00
GiteaMirror commented 2026-04-12 14:49:42 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Aug 11, 2024):

Hi gileneusz
I don't see a lower memory usage for the new Llama3.1:405b models. I think it's more performance and bug fixing.
Can you close this issue as the new models are in Ollama librairy

<!-- gh-comment-id:2282912404 --> @igorschlum commented on GitHub (Aug 11, 2024): Hi [gileneusz](https://github.com/gileneusz) I don't see a lower memory usage for the new Llama3.1:405b models. I think it's more performance and bug fixing. Can you close this issue as the new models are in Ollama [librairy](https://ollama.com/library/llama3.1/tags)
GiteaMirror commented 2026-04-12 14:49:42 -05:00
Author
Owner
Copy Link

@FellowTraveler commented on GitHub (Aug 12, 2024):

Hi gileneusz I don't see a lower memory usage for the new Llama3.1:405b models.

If the number of KV heads was changed from 16 to 8 and vllm was able to get a 20% memory reduction, shouldn't Ollama see the same benefit from that?
Can you confirm that the new version is the one in Ollama now? Seems like the model was just updated in Ollama 2 hours ago so I assume so.

<!-- gh-comment-id:2282934018 --> @FellowTraveler commented on GitHub (Aug 12, 2024): > Hi [gileneusz](https://github.com/gileneusz) I don't see a lower memory usage for the new Llama3.1:405b models. If the number of KV heads was changed from 16 to 8 and vllm was able to get a 20% memory reduction, shouldn't Ollama see the same benefit from that? Can you confirm that the new version is the one in Ollama now? Seems like the model was just updated in Ollama 2 hours ago so I assume so.
GiteaMirror commented 2026-04-12 14:49:42 -05:00
Author
Owner
Copy Link

@gileneusz commented on GitHub (Aug 12, 2024):

thanks for updating the model, I'll close the issue now, I'll test the fixed model and see if the memory usage is lower....

<!-- gh-comment-id:2283013613 --> @gileneusz commented on GitHub (Aug 12, 2024): thanks for updating the model, I'll close the issue now, I'll test the fixed model and see if the memory usage is lower....
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 model
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#3952
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?