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 #502] Better Document /api/embeddings #25992

New Issue
Closed
opened 2026-04-22 01:52:34 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-22 01:52:34 -05:00
Owner
Copy Link

Originally created by @FairyTail2000 on GitHub (Sep 10, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/502

It's currently not very clear on how to use is and how it's relation is to the EMBED Modelfile instruction.

Can you enlighten me and/or update the wiki and examples?

Originally created by @FairyTail2000 on GitHub (Sep 10, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/502 It's currently not very clear on how to use is and how it's relation is to the EMBED Modelfile instruction. Can you enlighten me and/or update the wiki and examples?
GiteaMirror commented 2026-04-22 01:52:35 -05:00
Author
Owner
Copy Link

@saul-jb commented on GitHub (Sep 14, 2023):

It's currently not very clear on how to use is and how it's relation is to the EMBED Modelfile instruction.

I am only just learning about this myself but here is my findings.

The embedding is a way to convert generic content into a Vector representation (a list of numbers). This is useful because we can perform vector mathematics on these to compare them and how closely they relate together. You could compare a list of embeddings created from a long list of files to an embedding created from a prompt, then you could use the embedding the most closely matches and reverse lookup the file of relevance. This is essentially a semantic search.

The /api/embeddings will take a prompt and generate an embedding from it and return it to you. It is up to you to figure out what you want to do with it.

The EMBED Modelfile instruction appears create an embedding which allows it to use that data to when providing a result.

<!-- gh-comment-id:1718810645 --> @saul-jb commented on GitHub (Sep 14, 2023): > It's currently not very clear on how to use is and how it's relation is to the EMBED Modelfile instruction. I am only just learning about this myself but here is my findings. The embedding is a way to convert generic content into a Vector representation (a list of numbers). This is useful because we can perform vector mathematics on these to compare them and how closely they relate together. You could compare a list of embeddings created from a long list of files to an embedding created from a prompt, then you could use the embedding the most closely matches and reverse lookup the file of relevance. This is essentially a semantic search. The /api/embeddings will take a prompt and generate an embedding from it and return it to you. It is up to you to figure out what you want to do with it. The EMBED Modelfile instruction appears create an embedding which allows it to use that data to when providing a result.
GiteaMirror commented 2026-04-22 01:52:36 -05:00
Author
Owner
Copy Link

@FairyTail2000 commented on GitHub (Sep 14, 2023):

Thanks for your explanation. It makes sense now. Maybe @jmorganca can cover this in the documentation

<!-- gh-comment-id:1719032847 --> @FairyTail2000 commented on GitHub (Sep 14, 2023): Thanks for your explanation. It makes sense now. Maybe @jmorganca can cover this in the documentation
GiteaMirror commented 2026-04-22 01:52:36 -05:00
Author
Owner
Copy Link

@fmnxl commented on GitHub (Sep 26, 2023):

So the embeddings won't be searched automatically when running inference with the model?

<!-- gh-comment-id:1735197739 --> @fmnxl commented on GitHub (Sep 26, 2023): So the embeddings won't be searched automatically when running inference with the model?
GiteaMirror commented 2026-04-22 01:52:36 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Sep 26, 2023):

It's not well documented since there is still a lot of work to do around embeddings, but you can automatically add the top 3 closest embeddings to a prompt using the {{ .Embed }} variable in your prompt template.

Here is an example Modelfile:

FROM llama2
EMBED <file path>.txt
TEMPLATE """
[INST] <<SYS>>You are a librarian that can answer questions about historical events with detail and accuracy.<</SYS>>

{{- if .Embed }}If this information is relevant use it to answer the prompt, otherwise ignore it: {{ .Embed }}
Prompt: {{- end }}{{ .Prompt }} [/INST]
<!-- gh-comment-id:1735635352 --> @BruceMacD commented on GitHub (Sep 26, 2023): It's not well documented since there is still a lot of work to do around embeddings, but you can automatically add the top 3 closest embeddings to a prompt using the `{{ .Embed }}` variable in your prompt template. Here is an example Modelfile: ``` FROM llama2 EMBED <file path>.txt TEMPLATE """ [INST] <<SYS>>You are a librarian that can answer questions about historical events with detail and accuracy.<</SYS>> {{- if .Embed }}If this information is relevant use it to answer the prompt, otherwise ignore it: {{ .Embed }} Prompt: {{- end }}{{ .Prompt }} [/INST] ```
GiteaMirror commented 2026-04-22 01:52:37 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Oct 26, 2023):

Hi @FairyTail2000, for documentation on /api/embeddings, check out https://github.com/jmorganca/ollama/blob/main/docs/api.md#generate-embeddings

In terms of using these embeddings, check out this guide by @simonw which is a great guide on what embeddings are and how to use them.

<!-- gh-comment-id:1780224430 --> @jmorganca commented on GitHub (Oct 26, 2023): Hi @FairyTail2000, for documentation on `/api/embeddings`, check out https://github.com/jmorganca/ollama/blob/main/docs/api.md#generate-embeddings In terms of using these embeddings, check out [this guide](https://simonwillison.net/2023/Oct/23/embeddings) by @simonw which is a great guide on what embeddings are and how to use them.
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#25992
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?