github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 08:59:13 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #10674] [MERGED] Multi-modal worst case graph #60016

New Issue
Closed
opened 2026-04-29 14:56:14 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-29 14:56:14 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/10674
Author: @jessegross
Created: 5/12/2025
Status: ✅ Merged
Merged: 5/15/2025
Merged by: @jessegross

Base: main ← Head: jessegross/worst-multimodal

📝 Commits (3)

  • 0f75589 ollamarunner: Base cached tokens on current prompt
  • 6c326f8 ollamarunner: Separate text and multimodal graphs
  • 7bd9f25 ollamarunner: Multi-modal worst case graph

📊 Changes

16 files changed (+319 additions, -190 deletions)

View changed files

📝 model/input/input.go (+20 -6)
📝 model/model.go (+5 -4)
📝 model/models/gemma3/model.go (+7 -7)
📝 model/models/gemma3/model_text.go (+1 -1)
📝 model/models/llama4/model.go (+53 -45)
📝 model/models/llama4/model_text.go (+1 -6)
📝 model/models/mistral3/model.go (+8 -33)
📝 model/models/mistral3/model_text.go (+1 -17)
📝 model/models/mllama/model.go (+5 -3)
📝 model/models/qwen25vl/model.go (+5 -34)
📝 model/models/qwen25vl/model_text.go (+1 -6)
📝 runner/llamarunner/cache.go (+1 -1)
📝 runner/ollamarunner/cache.go (+1 -1)
📝 runner/ollamarunner/cache_test.go (+6 -11)
➕ runner/ollamarunner/multimodal.go (+116 -0)
📝 runner/ollamarunner/runner.go (+88 -15)

📄 Description

We currently preallocate compute graph memory for the worst case batch of text tokens. This adds support for doing the same for images.

Note that image models are more complicated than text models in how they process their inputs so there may be cases where this approach isn't completely generic for all models. It covers all currently supported models though.

🔄 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/10674 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 5/12/2025 **Status:** ✅ Merged **Merged:** 5/15/2025 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/worst-multimodal` --- ### 📝 Commits (3) - [`0f75589`](https://github.com/ollama/ollama/commit/0f75589ee3e59ed9b0e24182db6e67336ea6fbf7) ollamarunner: Base cached tokens on current prompt - [`6c326f8`](https://github.com/ollama/ollama/commit/6c326f87d9bf81c9bafa68e14f92bdae98ff0a45) ollamarunner: Separate text and multimodal graphs - [`7bd9f25`](https://github.com/ollama/ollama/commit/7bd9f25cb2319d1c8da993ac62f7668f39127c54) ollamarunner: Multi-modal worst case graph ### 📊 Changes **16 files changed** (+319 additions, -190 deletions) <details> <summary>View changed files</summary> 📝 `model/input/input.go` (+20 -6) 📝 `model/model.go` (+5 -4) 📝 `model/models/gemma3/model.go` (+7 -7) 📝 `model/models/gemma3/model_text.go` (+1 -1) 📝 `model/models/llama4/model.go` (+53 -45) 📝 `model/models/llama4/model_text.go` (+1 -6) 📝 `model/models/mistral3/model.go` (+8 -33) 📝 `model/models/mistral3/model_text.go` (+1 -17) 📝 `model/models/mllama/model.go` (+5 -3) 📝 `model/models/qwen25vl/model.go` (+5 -34) 📝 `model/models/qwen25vl/model_text.go` (+1 -6) 📝 `runner/llamarunner/cache.go` (+1 -1) 📝 `runner/ollamarunner/cache.go` (+1 -1) 📝 `runner/ollamarunner/cache_test.go` (+6 -11) ➕ `runner/ollamarunner/multimodal.go` (+116 -0) 📝 `runner/ollamarunner/runner.go` (+88 -15) </details> ### 📄 Description We currently preallocate compute graph memory for the worst case batch of text tokens. This adds support for doing the same for images. Note that image models are more complicated than text models in how they process their inputs so there may be cases where this approach isn't completely generic for all models. It covers all currently supported models though. --- <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 14:56:14 -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#60016
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?