github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #15356] Gemma4 takes up too much space #56337

New Issue
Closed
opened 2026-04-29 10:40:02 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-29 10:40:02 -05:00
Owner
Copy Link

Originally created by @chllei on GitHub (Apr 6, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15356

In Google's AI Edge Gallery app, the E2B model requires only about 2GB of space, whereas the E2B model in Ollama needs 7.2GB. This substantial memory footprint is highly unsuitable for laptops with 16GB of RAM and no dedicated GPU, yet the model is purportedly designed for low-end devices, including Raspberry Pis.

Originally created by @chllei on GitHub (Apr 6, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15356 In Google's AI Edge Gallery app, the E2B model requires only about 2GB of space, whereas the E2B model in Ollama needs 7.2GB. This substantial memory footprint is highly unsuitable for laptops with 16GB of RAM and no dedicated GPU, yet the model is purportedly designed for low-end devices, including Raspberry Pis.
GiteaMirror added the model label 2026-04-29 10:40:03 -05:00
GiteaMirror commented 2026-04-29 10:40:26 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Apr 6, 2026):

The model as released by Google is 10G. The version used in the Gallery app is tuned for the LiteRT-LM framework and uses a more heavily quantized version of model. LiteRT-LM is specifcally designed for edge devices and has Raspberry pi support, if the environment is resource constrained then it may be a better framework than the more general purpose ollama.

<!-- gh-comment-id:4191479266 --> @rick-github commented on GitHub (Apr 6, 2026): The model as released by Google is [10G](https://huggingface.co/google/gemma-4-E2B-it/tree/main). The version used in the Gallery app is tuned for the [LiteRT-LM framework](https://ai.google.dev/edge/litert-lm/overview#featured_model_gemma-4-e2b) and uses a more heavily [quantized](https://huggingface.co/litert-community/gemma-4-E2B-it-litert-lm#:~:text=It%20uses%20the%20Gemma%20quantization%20scheme%20that%20employs%20a%20mixture%20of%202bit%2C%204bit%20and%208bit%20weights.) version of model. LiteRT-LM is specifcally designed for edge devices and has [Raspberry pi](https://github.com/google-ai-edge/LiteRT-LM#:~:text=Windows%20(WSL)%20or-,Raspberry%20Pi,-with%20the%20LiteRT) support, if the environment is resource constrained then it may be a better framework than the more general purpose ollama.
GiteaMirror commented 2026-04-29 10:40:27 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Apr 7, 2026):

The audio and vision tensors take up a substantial amount of space. We could release a version which doesn't include those, but no plans to do that yet.

<!-- gh-comment-id:4202460461 --> @pdevine commented on GitHub (Apr 7, 2026): The audio and vision tensors take up a substantial amount of space. We could release a version which doesn't include those, but no plans to do that yet.
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#56337
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?