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

[PR #10162] [CLOSED] create: check architecture rather than vision.block_count when importing GGUF #59859

New Issue
Closed
opened 2026-04-29 14:47:00 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-29 14:47:00 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/10162
Author: @rick-github
Created: 4/7/2025
Status: ❌ Closed

Base: main ← Head: projector

📝 Commits (1)

  • 51d2494 create: check arch rather than guess type from vision.block_count

📊 Changes

1 file changed (+1 additions, -1 deletions)

View changed files

📝 server/create.go (+1 -1)

📄 Description

TLDR: model files that have a vision component are mis-identified as projectors.

When importing a GGUF file, the type of the file is guessed from the presence of a vision.block_count KV entry. For models that fuse text and image weights into a single file (eg gemma3), this results in the file being classified as image.projector rather than image.model.

The only supported architectures for external projectors are currently clip and mllama, so use that as criteria rather than vision.block_count.

Fixes: #10036
Fixes: #10121
Fixes: #10719

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/10162 **Author:** [@rick-github](https://github.com/rick-github) **Created:** 4/7/2025 **Status:** ❌ Closed **Base:** `main` ← **Head:** `projector` --- ### 📝 Commits (1) - [`51d2494`](https://github.com/ollama/ollama/commit/51d249444e39e349202da0b111d560f55932b28d) create: check arch rather than guess type from vision.block_count ### 📊 Changes **1 file changed** (+1 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `server/create.go` (+1 -1) </details> ### 📄 Description TLDR: model files that have a vision component are mis-identified as projectors. When importing a GGUF file, the type of the file is guessed from the presence of a vision.block_count KV entry. For models that fuse text and image weights into a single file (eg gemma3), this results in the file being classified as image.projector rather than image.model. The only supported architectures for external projectors are currently [clip and mllama](https://github.com/ollama/ollama/blob/0f3f9e353df96d4cfc40ac19114c782a57fe30f5/runner/llamarunner/image.go#L36), so use that as criteria rather than vision.block_count. Fixes: #10036 Fixes: #10121 Fixes: #10719 --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-04-29 14:47:00 -05:00
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 pull-request
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#59859
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?