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 #5157] Update llama.cpp to support qwen2-57B-A14B pls #49762

New Issue
Closed
opened 2026-04-28 12:52:53 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-28 12:52:53 -05:00
Owner
Copy Link

Originally created by @CoreJa on GitHub (Jun 20, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5157

Originally assigned to: @dhiltgen on GitHub.

What is the issue?

llama.cpp is able to support qwen2-57B-A14B moe model now via https://github.com/ggerganov/llama.cpp/pull/7835. Please help update submodule llama.cpp to a newer version. Otherwise i quants of this model(I tried iq4_xs) would end up with a CUDA error: CUBLAS_STATUS_NOT_INITIALIZED.

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

v0.1.44

Originally created by @CoreJa on GitHub (Jun 20, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5157 Originally assigned to: @dhiltgen on GitHub. ### What is the issue? llama.cpp is able to support qwen2-57B-A14B moe model now via https://github.com/ggerganov/llama.cpp/pull/7835. Please help update submodule llama.cpp to a newer version. Otherwise i quants of this model(I tried iq4_xs) would end up with a `CUDA error: CUBLAS_STATUS_NOT_INITIALIZED`. ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version v0.1.44
GiteaMirror added the bug label 2026-04-28 12:52:53 -05:00
GiteaMirror commented 2026-04-28 12:52:55 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Jun 20, 2024):

This landed just after we updated for 0.1.45. We'll pick this up on our next update ~next week.

<!-- gh-comment-id:2180960040 --> @dhiltgen commented on GitHub (Jun 20, 2024): This landed just after we updated for 0.1.45. We'll pick this up on our next update ~next week.
GiteaMirror commented 2026-04-28 12:52:55 -05:00
Author
Owner
Copy Link

@CoreJa commented on GitHub (Jun 21, 2024):

Thx for your quick response, I'll close this issue when it releases. Appreciate your help!

<!-- gh-comment-id:2181847795 --> @CoreJa commented on GitHub (Jun 21, 2024): Thx for your quick response, I'll close this issue when it releases. Appreciate your help!
GiteaMirror commented 2026-04-28 12:52:55 -05:00
Author
Owner
Copy Link

@bunnyfu commented on GitHub (Jun 23, 2024):

Also something to keep in mind with the Qwen2 57b MoE model is that it breaks if you raise the batch above 256. Ollama tries to default to batches of 512. I have been manually setting in llama.cpp -ubatch 256 to make sure to avoid this error on some configs (macs): GGML_ASSERT: ggml-metal.m:1857: dst_rows <= 2048

More info here: https://github.com/ggerganov/llama.cpp/issues/7652

<!-- gh-comment-id:2184919158 --> @bunnyfu commented on GitHub (Jun 23, 2024): Also something to keep in mind with the Qwen2 57b MoE model is that it breaks if you raise the batch above 256. Ollama tries to default to batches of 512. I have been manually setting in llama.cpp -ubatch 256 to make sure to avoid this error on some configs (macs): `GGML_ASSERT: ggml-metal.m:1857: dst_rows <= 2048 ` More info here: https://github.com/ggerganov/llama.cpp/issues/7652
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 bug
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#49762
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?