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

[GH-ISSUE #4574] Error Loading Phi Medium #2869

New Issue
Open
opened 2026-04-12 13:13:12 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-12 13:13:12 -05:00
Owner
Copy Link

Originally created by @aneesha on GitHub (May 22, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4574

What is the issue?

ollama run phi3:medium
downloads the model but then shows this error:
Error: exception error loading model architecture: unknown model architecture: 'phi3'

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.1.29

Originally created by @aneesha on GitHub (May 22, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4574 ### What is the issue? ollama run phi3:medium downloads the model but then shows this error: Error: exception error loading model architecture: unknown model architecture: 'phi3' ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.1.29
GiteaMirror added the bug label 2026-04-12 13:13:12 -05:00
GiteaMirror commented 2026-04-12 13:13:14 -05:00
Author
Owner
Copy Link

@coder543 commented on GitHub (May 22, 2024):

You need the pre-release version: https://github.com/ollama/ollama/releases/tag/v0.1.39-rc1

Phi-3 Medium isn't supported on the regular releases yet.

<!-- gh-comment-id:2124757916 --> @coder543 commented on GitHub (May 22, 2024): You need the pre-release version: https://github.com/ollama/ollama/releases/tag/v0.1.39-rc1 Phi-3 Medium isn't supported on the regular releases yet.
GiteaMirror commented 2026-04-12 13:13:14 -05:00
Author
Owner
Copy Link

@napa3um commented on GitHub (May 23, 2024):

0.1.39-rc1 load medium model Q_6, Q_4 from https://ollama.com/library/phi3 (Apple M2-Max), but:

request: привет
response: Отлично, что вы добрались! Чем я могу вам помочь?

Instruction 2 (much more difficult with at least 5 additional constraints):

` salutations  `  

Create a complex Russian phrase that includes:
- A polite greeting appropriate for the afternoon.
- Mention of two authors, "Толстой" and "Достоевский", in reference to their literary contributions.
- An inquiry about an upcoming cultural event related to classical literature.
- Use at least one idiomatic expression that is commonly used in Russian conversation.
- Express interest in discussing the philosophical implications of literature on modern society.
- The phrase should be written as if it were part of a letter and formatted accordingly, complete with formal closing salutations.

And so almost always some unnecessary instructions, sometimes endless.

<!-- gh-comment-id:2127189379 --> @napa3um commented on GitHub (May 23, 2024): 0.1.39-rc1 load medium model Q_6, Q_4 from https://ollama.com/library/phi3 (Apple M2-Max), but: ----------------------------- **request:** привет **response:** Отлично, что вы добрались! Чем я могу вам помочь? ``` Instruction 2 (much more difficult with at least 5 additional constraints): ` salutations ` Create a complex Russian phrase that includes: - A polite greeting appropriate for the afternoon. - Mention of two authors, "Толстой" and "Достоевский", in reference to their literary contributions. - An inquiry about an upcoming cultural event related to classical literature. - Use at least one idiomatic expression that is commonly used in Russian conversation. - Express interest in discussing the philosophical implications of literature on modern society. - The phrase should be written as if it were part of a letter and formatted accordingly, complete with formal closing salutations. ``` ----------------------------- And so almost always some unnecessary instructions, sometimes endless.
GiteaMirror commented 2026-04-12 13:13:15 -05:00
Author
Owner
Copy Link

@dstaicova commented on GitHub (May 24, 2024):

Is this going to be fixed soon? It's rather disappointing to wait for 8GB to download just to see an Error message. Besides, maybe for future such situations, it's best to have a separate error for not supported models so that people don't scratch their heads. (P.S. I'm on Manjaro, so who knows when the new ollama will be updated)

<!-- gh-comment-id:2129829842 --> @dstaicova commented on GitHub (May 24, 2024): Is this going to be fixed soon? It's rather disappointing to wait for 8GB to download just to see an Error message. Besides, maybe for future such situations, it's best to have a separate error for not supported models so that people don't scratch their heads. (P.S. I'm on Manjaro, so who knows when the new ollama will be updated)
GiteaMirror commented 2026-04-12 13:13:15 -05:00
Author
Owner
Copy Link

@coder543 commented on GitHub (May 24, 2024):

You can download the pre-release version at any time if you want to test out Phi3 Medium. As someone outside the project, I will say that I usually see the pre-release versions convert into the new release within a couple of days, so it probably won't be long now.

<!-- gh-comment-id:2129841283 --> @coder543 commented on GitHub (May 24, 2024): You can download the [pre-release version](https://github.com/ollama/ollama/releases/tag/v0.1.39) at any time if you want to test out Phi3 Medium. As someone outside the project, I will say that I usually see the pre-release versions convert into the new release within a couple of days, so it probably won't be long now.
GiteaMirror commented 2026-04-12 13:13:16 -05:00
Author
Owner
Copy Link

@dstaicova commented on GitHub (May 24, 2024):

You can download the pre-release version at any time if you want to test out Phi3 Medium. As someone outside the project, I will say that I usually see the pre-release versions convert into the new release within a couple of days, so it probably won't be long now.

Oh, ok, then. Let's hope Arch would import it rather sooner than later :) I'd rather not install the pre-release, because I already have way too many conda envs.

<!-- gh-comment-id:2130314710 --> @dstaicova commented on GitHub (May 24, 2024): > You can download the [pre-release version](https://github.com/ollama/ollama/releases/tag/v0.1.39) at any time if you want to test out Phi3 Medium. As someone outside the project, I will say that I usually see the pre-release versions convert into the new release within a couple of days, so it probably won't be long now. Oh, ok, then. Let's hope Arch would import it rather sooner than later :) I'd rather not install the pre-release, because I already have way too many conda envs.
GiteaMirror commented 2026-04-12 13:13:18 -05:00
Author
Owner
Copy Link

@napa3um commented on GitHub (May 25, 2024):

apple m2-max 32gb: phi3-medium on 0.1.39 is worked, but num_ctx 128000 in modelfile leads to loading cpu 400% and freeze system instead response :(

<!-- gh-comment-id:2131379741 --> @napa3um commented on GitHub (May 25, 2024): apple m2-max 32gb: phi3-medium on 0.1.39 is worked, but num_ctx 128000 in modelfile leads to loading cpu 400% and freeze system instead response :(
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#2869
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?