github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #4447] NumCtx can't change, just 2048 #64815

New Issue
Closed
opened 2026-05-03 18:52:56 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-05-03 18:52:56 -05:00
Owner
Copy Link

Originally created by @jianwen-wang on GitHub (May 15, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4447

What is the issue?

When using the LLaMA 3 model with OpenAI’s /v1/chat/completions API, it was discovered that the request message can’t exceed 2k tokens, while the LLaMA 3 8B model inherently supports up to 8K tokens.

Upon examining the code, it was found that in api/types.go at line 475, DefaultOptions contains a fixed default value of 2048 for NumCtx, and there is no available option to modify this default configuration.

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.1.37

Originally created by @jianwen-wang on GitHub (May 15, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4447 ### What is the issue? When using the LLaMA 3 model with OpenAI’s /v1/chat/completions API, it was discovered that the request message can’t exceed 2k tokens, while the LLaMA 3 8B model inherently supports up to 8K tokens. Upon examining the code, it was found that in api/types.go at line 475, DefaultOptions contains a fixed default value of 2048 for NumCtx, and there is no available option to modify this default configuration. ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.1.37
GiteaMirror added the bug label 2026-05-03 18:52:56 -05:00
GiteaMirror commented 2026-05-03 18:53:04 -05:00
Author
Owner
Copy Link

@mxyng commented on GitHub (May 15, 2024):

The OpenAI API spec does not provide a mechanism to set context length. However, if you use the Ollama API, the context length can be set a few ways:

  1. Statically for a model through the Modelfile, e.g. PARAMETER num_ctx 8192
  2. Dynamically during runtime with API options, e.g. {"model": "llama3", "options": {"num_ctx": 8192}}
  3. Dynamically during runtime with client library options, e.g. for Python ollama.chat(model="llama3", options={"num_ctx": 8192})
<!-- gh-comment-id:2112784702 --> @mxyng commented on GitHub (May 15, 2024): The OpenAI API [spec](https://platform.openai.com/docs/api-reference/chat/create) does not provide a mechanism to set context length. However, if you use the Ollama API, the context length can be set a few ways: 1. Statically for a model through the Modelfile, e.g. `PARAMETER num_ctx 8192` 2. Dynamically during runtime with API options, e.g. `{"model": "llama3", "options": {"num_ctx": 8192}}` 3. Dynamically during runtime with client library options, e.g. for Python `ollama.chat(model="llama3", options={"num_ctx": 8192})`
GiteaMirror commented 2026-05-03 18:53:05 -05:00
Author
Owner
Copy Link

@wszgrcy commented on GitHub (Aug 2, 2024):

The OpenAI API spec does not provide a mechanism to set context length. However, if you use the Ollama API, the context length can be set a few ways:

  1. Statically for a model through the Modelfile, e.g. PARAMETER num_ctx 8192
  2. Dynamically during runtime with API options, e.g. {"model": "llama3", "options": {"num_ctx": 8192}}
  3. Dynamically during runtime with client library options, e.g. for Python ollama.chat(model="llama3", options={"num_ctx": 8192})

I use num_ctx by /v1/chat/completions but not work
Does this parameter only support the '/app/generate' interface?

<!-- gh-comment-id:2264425558 --> @wszgrcy commented on GitHub (Aug 2, 2024): > The OpenAI API [spec](https://platform.openai.com/docs/api-reference/chat/create) does not provide a mechanism to set context length. However, if you use the Ollama API, the context length can be set a few ways: > > 1. Statically for a model through the Modelfile, e.g. `PARAMETER num_ctx 8192` > 2. Dynamically during runtime with API options, e.g. `{"model": "llama3", "options": {"num_ctx": 8192}}` > 3. Dynamically during runtime with client library options, e.g. for Python `ollama.chat(model="llama3", options={"num_ctx": 8192})` I use `num_ctx` by `/v1/chat/completions` but not work Does this parameter only support the '/app/generate' interface?
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#64815
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?