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 #4104] Unexpected embedding model response ("Embedding model" blogpost) #28311

New Issue
Closed
opened 2026-04-22 06:21:36 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-22 06:21:36 -05:00
Owner
Copy Link

Originally created by @asmith26 on GitHub (May 2, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4104

What is the issue?

Hi,

I'm running the example from https://ollama.com/blog/embedding-models, and when I look at the value of data = results['documents'][0][0] every time I get:

Llamas are vegetarians and have very efficient digestive systems

The blog implies the value it gets for data (based on the generated answer, it doesn't actually say):

Llamas are members of the camelid family meaning they're pretty closely related to vicuñas and camels

Not sure if anyone else gets this data output when running the example from the "Embedding model" blogpost? I've done some further analysis and it does appear that possibly ollama embeddings have degraded a little between v0.1.31 and v0.1.32
Thanks!

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.1.32

Originally created by @asmith26 on GitHub (May 2, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4104 ### What is the issue? Hi, I'm running the example from https://ollama.com/blog/embedding-models, and when I look at the value of `data = results['documents'][0][0]` every time I get: ``` Llamas are vegetarians and have very efficient digestive systems ``` The blog implies the value it gets for `data` (based on the generated answer, it doesn't actually say): ``` Llamas are members of the camelid family meaning they're pretty closely related to vicuñas and camels ``` Not sure if anyone else gets this `data` output when running the example from the "Embedding model" blogpost? I've done some further analysis and it does appear that possibly ollama embeddings have degraded a little between v0.1.31 and v0.1.32 Thanks! ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.1.32
GiteaMirror added the bug label 2026-04-22 06:21:36 -05:00
GiteaMirror commented 2026-04-22 06:21:39 -05:00
Author
Owner
Copy Link

@thinkverse commented on GitHub (May 2, 2024):

IDK if I'd call it unexpected. It is likely that LLama already has information related to lamas in its corpus and prioritizes that over the unrelated data in the prompt.

Trying the same example with gemma gives a more reasonable response given the provided data.

The provided text does not contain any information regarding the animals 
related to llamas, so I am unable to answer this question from the given context

If we for example switch to another embedding model like nomic-embed-text, which does grab more related data from our knowledge documents, gemma gives a better answer.

Llamas are related to vicuñas and camels, as they are all members of the camelid family.
<!-- gh-comment-id:2091559815 --> @thinkverse commented on GitHub (May 2, 2024): IDK if I'd call it unexpected. It is likely that LLama already has information related to lamas in its corpus and prioritizes that over the unrelated data in the prompt. Trying the same example with `gemma` gives a more reasonable response given the provided data. ```shell The provided text does not contain any information regarding the animals related to llamas, so I am unable to answer this question from the given context ``` If we for example switch to another embedding model like `nomic-embed-text`, which does grab more related data from our knowledge documents, `gemma` gives a better answer. ```shell Llamas are related to vicuñas and camels, as they are all members of the camelid family. ```
GiteaMirror commented 2026-04-22 06:21:40 -05:00
Author
Owner
Copy Link

@asmith26 commented on GitHub (May 2, 2024):

Thanks very much for your help @thinkverse! Can confirm when using the nomic-embed-text embedding model it returns the more expected information.

Just wondering, do you know if there is a useful way to determine which embedding models are better than others? Thanks again for any help!

<!-- gh-comment-id:2091601768 --> @asmith26 commented on GitHub (May 2, 2024): Thanks very much for your help @thinkverse! Can confirm when using the `nomic-embed-text` embedding model it returns the more expected information. Just wondering, do you know if there is a useful way to determine which embedding models are better than others? Thanks again for any help!
GiteaMirror commented 2026-04-22 06:21:42 -05:00
Author
Owner
Copy Link

@thinkverse commented on GitHub (May 2, 2024):

Just wondering, do you know if there is a useful way to determine which embedding models are better than others?

My experience with them is rather limited but for me. I use the tried and true way of trial and error. 👍 I like nomic-embed-text though, and haven't had any issues with it.

<!-- gh-comment-id:2091628980 --> @thinkverse commented on GitHub (May 2, 2024): > Just wondering, do you know if there is a useful way to determine which embedding models are better than others? My experience with them is rather limited but for me. I use the tried and true way of trial and error. 👍 I like `nomic-embed-text` though, and haven't had any issues with it.
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#28311
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?