[GH-ISSUE #3119] Tensor token_embed.weight has wrong shape #27679

Closed
opened 2026-04-22 05:12:53 -05:00 by GiteaMirror · 4 comments
Owner

Originally created by @BruceMacD on GitHub (Mar 13, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3119

image
split from #2753

Originally created by @BruceMacD on GitHub (Mar 13, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3119 ![image](https://github.com/ollama/ollama/assets/5853428/543d8714-16de-421f-b669-2afd9b87fe81) split from #2753
GiteaMirror added the bug label 2026-04-22 05:12:53 -05:00
Author
Owner

@BruceMacD commented on GitHub (Mar 13, 2024):

cc: @FotieMConstant, opened an issue specific to your problem here

<!-- gh-comment-id:1995206000 --> @BruceMacD commented on GitHub (Mar 13, 2024): cc: @FotieMConstant, opened an issue specific to your problem here
Author
Owner

@mxyng commented on GitHub (Mar 13, 2024):

this should be fixed by #3044 assuming the model was created using the convert from safetensors feature

<!-- gh-comment-id:1995271852 --> @mxyng commented on GitHub (Mar 13, 2024): this should be fixed by #3044 assuming the model was created using the convert from safetensors feature
Author
Owner

@FotieMConstant commented on GitHub (Mar 13, 2024):

hey @mxyng could you please explain what you mean by "assuming the model was created using the convert from safetensors feature" or how it can be done(maybe an example could help), i am not so experienced in this and this is my first time creating a fine-tune from Llama2-7b.

In the main time, i completely uninstalled ollama from my mac and did a clean install converted and quantized the model with the following instructions at import.md. create a model file and tried to run it again but it still give me the same error.

Any idea how i can fix it?
I'd really appreciate some help:)
@BruceMacD

<!-- gh-comment-id:1995692210 --> @FotieMConstant commented on GitHub (Mar 13, 2024): hey @mxyng could you please explain what you mean by "**assuming the model was created using the convert from safetensors feature**" or how it can be done(maybe an example could help), i am not so experienced in this and this is my first time creating a fine-tune from Llama2-7b. In the main time, i completely uninstalled ollama from my mac and did a clean install converted and quantized the model with the following instructions at [import.md](https://github.com/ollama/ollama/blob/main/docs/import.md). create a model file and tried to run it again but it still give me the same error. Any idea how i can fix it? I'd really appreciate some help:) @BruceMacD
Author
Owner

@FotieMConstant commented on GitHub (Mar 13, 2024):

been stuck on this for days:(

<!-- gh-comment-id:1995698343 --> @FotieMConstant commented on GitHub (Mar 13, 2024): been stuck on this for days:(
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#27679