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 #9969] Some Questions about Using Embedding Models in Ollama #32289

New Issue
Closed
opened 2026-04-22 13:24:57 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-22 13:24:57 -05:00
Owner
Copy Link

Originally created by @20246688 on GitHub (Mar 25, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9969

Hello! I've noticed that when using vector generation models on Ollama, the speed is really fast. Currently, I've found two links where I can learn how to use vector generation models on Ollama. They are https://github.com/ollama/ollama-python/blob/main/examples/embed.py and https://ollama.com/blog/embedding-models. But I'm also really eager to learn if there are any other parameters that I need to pay attention to when setting them up. Are there any specific details about parameter configuration that I should be aware of?

Originally created by @20246688 on GitHub (Mar 25, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9969 Hello! I've noticed that when using vector generation models on Ollama, the speed is really fast. Currently, I've found two links where I can learn how to use vector generation models on Ollama. They are https://github.com/ollama/ollama-python/blob/main/examples/embed.py and https://ollama.com/blog/embedding-models. But I'm also really eager to learn if there are any other parameters that I need to pay attention to when setting them up. Are there any specific details about parameter configuration that I should be aware of?
GiteaMirror commented 2026-04-22 13:24:59 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 26, 2025):

There's not much that needs to be tuned for embedding models. However, there is an outstanding issue regarding context size. TL;DR: explictly set num_ctx in the API call to the context size supported by the model. Setting it higher, or letting ollama use the default of 2048, can cause the runner to crash. The length of the text (chunk size) that you feed to the embedding model should be such that the number of tokens that the text is turned in to is less than the context size, otherwise you risk losing semantic content in the returned embedding.

<!-- gh-comment-id:2753011125 --> @rick-github commented on GitHub (Mar 26, 2025): There's not much that needs to be tuned for embedding models. However, there is an [outstanding issue](https://github.com/ollama/ollama/issues/7288) regarding context size. TL;DR: explictly set `num_ctx` in the API call to the context size supported by the model. Setting it higher, or letting ollama use the default of 2048, can cause the runner to crash. The length of the text (chunk size) that you feed to the embedding model should be such that the number of tokens that the text is turned in to is less than the context size, otherwise you risk losing semantic content in the returned embedding.
GiteaMirror commented 2026-04-22 13:24:59 -05:00
Author
Owner
Copy Link

@20246688 commented on GitHub (Mar 26, 2025):

There's not much that needs to be tuned for embedding models. However, there is an outstanding issue regarding context size. TL;DR: explictly set in the API call to the context size supported by the model. Setting it higher, or letting ollama use the default of 2048, can cause the runner to crash. The length of the text (chunk size) that you feed to the embedding model should be such that the number of tokens that the text is turned in to is less than the context size, otherwise you risk losing semantic content in the returned embedding.num_ctx

Thank you so much. I had a vague feeling that there would be context limitations before. I really appreciate your confirmation!

<!-- gh-comment-id:2753678893 --> @20246688 commented on GitHub (Mar 26, 2025): > There's not much that needs to be tuned for embedding models. However, there is an [outstanding issue](https://github.com/ollama/ollama/issues/7288) regarding context size. TL;DR: explictly set in the API call to the context size supported by the model. Setting it higher, or letting ollama use the default of 2048, can cause the runner to crash. The length of the text (chunk size) that you feed to the embedding model should be such that the number of tokens that the text is turned in to is less than the context size, otherwise you risk losing semantic content in the returned embedding.`num_ctx` Thank you so much. I had a vague feeling that there would be context limitations before. I really appreciate your confirmation!
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#32289
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?