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 #5409] [CLOSED] convert: only extract large files #37642

New Issue
Closed
opened 2026-04-22 22:19:07 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-22 22:19:07 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/5409
Author: @mxyng
Created: 7/1/2024
Status: ❌ Closed

Base: mxyng/convert2 ← Head: mxyng/convert-fs

📝 Commits (1)

  • 3ff2b70 convert: only extract large files

📊 Changes

10 files changed (+117 additions, -197 deletions)

View changed files

📝 convert/convert.go (+5 -6)
📝 convert/convert_test.go (+4 -3)
➕ convert/fs.go (+58 -0)
📝 convert/reader.go (+5 -5)
📝 convert/reader_safetensors.go (+14 -6)
📝 convert/reader_torch.go (+2 -1)
📝 convert/tokenizer.go (+11 -11)
📝 convert/tokenizer_spm.go (+4 -4)
📝 server/model.go (+14 -59)
📝 server/model_test.go (+0 -102)

📄 Description

many of the files needed during conversion don't need to be extracted and can be read directly from the zip. the only exception are the model weights. while it's possible to read these directly from the zip, the impact on performance is unacceptable (2m30s vs. 30s for gemma 2b). for these files, extract them when needed

anything larger than the limit (default 32 MiB) will be extracted to disk

the performance difference of this change (Mistral 7B ~= 53.828s) is negligible compared to the baseline (~= 54.23s) (n=1)

a follow up would be to remove the extracted file once they're no longer used so only one large file exists on disk at any given time

🔄 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/5409 **Author:** [@mxyng](https://github.com/mxyng) **Created:** 7/1/2024 **Status:** ❌ Closed **Base:** `mxyng/convert2` ← **Head:** `mxyng/convert-fs` --- ### 📝 Commits (1) - [`3ff2b70`](https://github.com/ollama/ollama/commit/3ff2b70af21413bc57397bfad45038419c9bebe8) convert: only extract large files ### 📊 Changes **10 files changed** (+117 additions, -197 deletions) <details> <summary>View changed files</summary> 📝 `convert/convert.go` (+5 -6) 📝 `convert/convert_test.go` (+4 -3) ➕ `convert/fs.go` (+58 -0) 📝 `convert/reader.go` (+5 -5) 📝 `convert/reader_safetensors.go` (+14 -6) 📝 `convert/reader_torch.go` (+2 -1) 📝 `convert/tokenizer.go` (+11 -11) 📝 `convert/tokenizer_spm.go` (+4 -4) 📝 `server/model.go` (+14 -59) 📝 `server/model_test.go` (+0 -102) </details> ### 📄 Description many of the files needed during conversion don't need to be extracted and can be read directly from the zip. the only exception are the model weights. while it's possible to read these directly from the zip, the impact on performance is unacceptable (2m30s vs. 30s for gemma 2b). for these files, extract them when needed anything larger than the limit (default 32 MiB) will be extracted to disk the performance difference of this change (Mistral 7B \~= 53.828s) is negligible compared to the baseline (\~= 54.23s) (n=1) a follow up would be to remove the extracted file once they're no longer used so only one large file exists on disk at any given time --- <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 2026-04-22 22:19:07 -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#37642
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?