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

[GH-ISSUE #10093] Error creating Ollama model from hugging face microsoft/phi4 #53129

New Issue
Closed
opened 2026-04-29 02:02:05 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-29 02:02:05 -05:00
Owner
Copy Link

Originally created by @nishtahir on GitHub (Apr 2, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/10093

What is the issue?

Ollama model fails to run when created using microsoft/phi-4 huggingface model with

Error: llama runner process has terminated: error loading model: check_tensor_dims: tensor 'rope_factors_long.weight' has wrong shape; expected    64, got     0,     1,     1,     1
llama_model_load_from_file_impl: failed to load model

Steps to reproduce

  1. Save the hugging face model output
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model = AutoModelForCausalLM.from_pretrained("microsoft/phi-4")
tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-4")

base_model.save_pretrained("model_output")
tokenizer.save_pretrained("model_output")
  1. Create a Modelfile in the model_output directory
FROM .
  1. Create an ollama model using the Modelfile
ollama create -f model_output/Modelfile microsoft/phi-4

The model is created successfully

...
converting model 
using existing layer sha256:540d5dbc018433acaadee51346944cb9b79f6dcb50c653a784a5c621061226c6 
using autodetected template chatml 
using existing layer sha256:f02dd72bb2423204352eabc5637b44d79d17f109fdb510a7c51455892aa2d216 
writing manifest 
success
  1. Attempt to run the model
ollama run microsoft/phi-4

This fails with Error: llama runner process has terminated: error loading model: check_tensor_dims: tensor 'rope_factors_long.weight' has wrong shape; expected 64, got 0, 1, 1, 1 llama_model_load_from_file_impl: failed to load model

Relevant log output

ollama show microsoft/phi-4

  Model
    architecture        phi3     
    parameters          14.7B    
    context length      16384    
    embedding length    5120     
    quantization        F16      

  Parameters
    stop    "<|im_start|>"    
    stop    "<|im_end|>"

OS

Linux

GPU

Nvidia

CPU

AMD, Intel

Ollama version

0.6.2

Originally created by @nishtahir on GitHub (Apr 2, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/10093 ### What is the issue? Ollama model fails to run when created using `microsoft/phi-4` huggingface model with ``` Error: llama runner process has terminated: error loading model: check_tensor_dims: tensor 'rope_factors_long.weight' has wrong shape; expected 64, got 0, 1, 1, 1 llama_model_load_from_file_impl: failed to load model ``` # Steps to reproduce 1. Save the hugging face model output ``` from transformers import AutoModelForCausalLM, AutoTokenizer base_model = AutoModelForCausalLM.from_pretrained("microsoft/phi-4") tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-4") base_model.save_pretrained("model_output") tokenizer.save_pretrained("model_output") ``` 2. Create a `Modelfile` in the `model_output` directory ``` FROM . ``` 3. Create an ollama model using the `Modelfile` ``` ollama create -f model_output/Modelfile microsoft/phi-4 ``` The model is created successfully ``` ... converting model using existing layer sha256:540d5dbc018433acaadee51346944cb9b79f6dcb50c653a784a5c621061226c6 using autodetected template chatml using existing layer sha256:f02dd72bb2423204352eabc5637b44d79d17f109fdb510a7c51455892aa2d216 writing manifest success ``` 4. Attempt to run the model ``` ollama run microsoft/phi-4 ``` This fails with `Error: llama runner process has terminated: error loading model: check_tensor_dims: tensor 'rope_factors_long.weight' has wrong shape; expected 64, got 0, 1, 1, 1 llama_model_load_from_file_impl: failed to load model` ### Relevant log output ```shell ollama show microsoft/phi-4 Model architecture phi3 parameters 14.7B context length 16384 embedding length 5120 quantization F16 Parameters stop "<|im_start|>" stop "<|im_end|>" ``` ### OS Linux ### GPU Nvidia ### CPU AMD, Intel ### Ollama version 0.6.2
GiteaMirror added the bug label 2026-04-29 02:02:05 -05:00
GiteaMirror commented 2026-04-29 02:02:07 -05:00
Author
Owner
Copy Link

@nishtahir commented on GitHub (Apr 2, 2025):

A workaround appears to be converting the model to GGUF using llamacpp and creating the ollama model that way.

llama-convert-hf-to-gguf ./model_output
ollama create -f model_output/Modelfile microsoft/phi-4
<!-- gh-comment-id:2773748379 --> @nishtahir commented on GitHub (Apr 2, 2025): A workaround appears to be converting the model to GGUF using `llamacpp` and creating the ollama model that way. ``` llama-convert-hf-to-gguf ./model_output ollama create -f model_output/Modelfile microsoft/phi-4 ```
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#53129
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?