github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-15 20:54:54 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #9178] Using local models with huggingface autotokenizer #52488

New Issue
Closed
opened 2026-04-28 23:26:15 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-28 23:26:15 -05:00
Owner
Copy Link

Originally created by @Bgfreek on GitHub (Feb 17, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9178

Hi everyone,

I am trying to run vanilj/phi-4-unsloth:latest which I downloaded using ollama via "ollama run (model name)".

I am running ollama on a Ubuntu distribution via WSL. I also mapped the network to a drive so now the drive that Ubuntu is running on is "Z:"

When I call the models I do it:

  1. LOCAL_MODEL_BASE_PATH = r"Z:/usr/....../manfests/registry.ollama.ai/" (I do this to provide the default path to all the models)

  2. DEFAULT_SUMMARIZATION_MODEL_ID = r"vanilj/phi-4-unsloth/latest" (I duse different models for different tasks by adding only the endigs to access models where needed)

Now I am trying to access the local models but I get two errors:

  1. Incorrect path_or_model_id: 'vanilj/phi-4-unsloth:latest'. Please provide either the path to a local folder or the repo_id of a model on the Hub.

  2. huggingface_hub.errors.HFValidationError: Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'vanilj/phi-4-unsloth:latest'.

Thank you for your time in advance!

Originally created by @Bgfreek on GitHub (Feb 17, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9178 Hi everyone, I am trying to run vanilj/phi-4-unsloth:latest which I downloaded using ollama via "ollama run (model name)". I am running ollama on a Ubuntu distribution via WSL. I also mapped the network to a drive so now the drive that Ubuntu is running on is "Z:" When I call the models I do it: 1. LOCAL_MODEL_BASE_PATH = r"Z:/usr/....../manfests/registry.ollama.ai/" (I do this to provide the default path to all the models) 2. DEFAULT_SUMMARIZATION_MODEL_ID = r"vanilj/phi-4-unsloth/latest" (I duse different models for different tasks by adding only the endigs to access models where needed) Now I am trying to access the local models but I get two errors: 1. Incorrect path_or_model_id: 'vanilj/phi-4-unsloth:latest'. Please provide either the path to a local folder or the repo_id of a model on the Hub. 2. huggingface_hub.errors.HFValidationError: Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'vanilj/phi-4-unsloth:latest'. Thank you for your time in advance!
GiteaMirror commented 2026-04-28 23:26:16 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 17, 2025):

ollama doesn't save the models in the same format as HF. If you examine the manifest (.../manfests/registry.ollama.ai/vanilj/phi-4-unsloth/latest) you will see the blob identifier. If it was going to work, you need to use the path to that file (.../blobs/sha256-xxxxxxxxxxx). I don't know if that will be enough because the model will be quantized and your framework may not accept it.

<!-- gh-comment-id:2663976087 --> @rick-github commented on GitHub (Feb 17, 2025): ollama doesn't save the models in the same format as HF. If you examine the manifest (`.../manfests/registry.ollama.ai/vanilj/phi-4-unsloth/latest`) you will see the blob identifier. If it was going to work, you need to use the path to that file (`.../blobs/sha256-xxxxxxxxxxx`). I don't know if that will be enough because the model will be quantized and your framework may not accept it.
GiteaMirror commented 2026-04-28 23:26:17 -05:00
Author
Owner
Copy Link

@Bgfreek commented on GitHub (Feb 18, 2025):

Thank you very much for your reply!

Indeed, agreed. I was wondering if there was any way to do this but it seems that the only way to accomplish the goal is to host a model from huggingface locally and only call the llm for text/prompt based work.

<!-- gh-comment-id:2665710013 --> @Bgfreek commented on GitHub (Feb 18, 2025): Thank you very much for your reply! Indeed, agreed. I was wondering if there was any way to do this but it seems that the only way to accomplish the goal is to host a model from huggingface locally and only call the llm for text/prompt based work.
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
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#52488
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?