github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #11036] gemma3:27b unable to get full 128k context #7279

New Issue
Closed
opened 2026-04-12 19:19:43 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 19:19:43 -05:00
Owner
Copy Link

Originally created by @pavanrajkg04 on GitHub (Jun 10, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11036

What is the issue?

i tried with the gemma3:27b model but I am unable to get the full 128k context as I am working on chatbot that takes the SQL data. i am unable to deliver what I am expecting. i need help on this

Relevant log output

OS

macOS, Linux

GPU

Nvidia

CPU

No response

Ollama version

v0.9.1

Originally created by @pavanrajkg04 on GitHub (Jun 10, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11036 ### What is the issue? i tried with the gemma3:27b model but I am unable to get the full 128k context as I am working on chatbot that takes the SQL data. i am unable to deliver what I am expecting. i need help on this ### Relevant log output ```shell ``` ### OS macOS, Linux ### GPU Nvidia ### CPU _No response_ ### Ollama version v0.9.1
GiteaMirror added the bug label 2026-04-12 19:19:43 -05:00
GiteaMirror commented 2026-04-12 19:19:45 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jun 10, 2025):

but I am unable to get the full 128k context

What does this mean? Your client can't send 128k tokens of data? The model doesn't load because the context is too big? The model doesn't generate tokens when the context is 128k?

<!-- gh-comment-id:2960242556 --> @rick-github commented on GitHub (Jun 10, 2025): > but I am unable to get the full 128k context What does this mean? Your client can't send 128k tokens of data? The model doesn't load because the context is too big? The model doesn't generate tokens when the context is 128k?
GiteaMirror commented 2026-04-12 19:19:45 -05:00
Author
Owner
Copy Link

@duck-5 commented on GitHub (Jun 10, 2025):

Please provide more context, the GPU model, and if possible the Modelfile and the request json.

<!-- gh-comment-id:2960277015 --> @duck-5 commented on GitHub (Jun 10, 2025): Please provide more context, the GPU model, and if possible the Modelfile and the request json.
GiteaMirror commented 2026-04-12 19:19:45 -05:00
Author
Owner
Copy Link

@pavanrajkg04 commented on GitHub (Jun 11, 2025):

I found the solution for this. Ollama doesn't directly support 128k context; it maxes out at 32k context.
using this command we can increase the context length:
export OLLAMA_CONTEXT_LENGTH=131072

<!-- gh-comment-id:2961262386 --> @pavanrajkg04 commented on GitHub (Jun 11, 2025): I found the solution for this. Ollama doesn't directly support 128k context; it maxes out at 32k context. using this command we can increase the context length: export OLLAMA_CONTEXT_LENGTH=131072
GiteaMirror commented 2026-04-12 19:19:46 -05:00
Author
Owner
Copy Link

@pavanrajkg04 commented on GitHub (Jun 11, 2025):

but I am unable to get the full 128k context

What does this mean? Your client can't send 128k tokens of data? The model doesn't load because the context is too big? The model doesn't generate tokens when the context is 128k?

The problem was:

  1. I used a RAG-based SQL generator. While a user asks a question, the LLM has to generate the SQL using the available schema. As my schema was very large, the model context would overflow, and I would get hallucinated SQL.
  2. Once generated, the model would also get 1000+ rows of data and it was unable to generate the correct summary.

i searched on the web and found that Ollama doesn't directly support 128k context length, it max out to 32k and using the following command we can increase it "export OLLAMA_CONTEXT_LENGTH=131072"

<!-- gh-comment-id:2961274718 --> @pavanrajkg04 commented on GitHub (Jun 11, 2025): > > but I am unable to get the full 128k context > > What does this mean? Your client can't send 128k tokens of data? The model doesn't load because the context is too big? The model doesn't generate tokens when the context is 128k? The problem was: 1. I used a RAG-based SQL generator. While a user asks a question, the LLM has to generate the SQL using the available schema. As my schema was very large, the model context would overflow, and I would get hallucinated SQL. 2. Once generated, the model would also get 1000+ rows of data and it was unable to generate the correct summary. i searched on the web and found that Ollama doesn't directly support 128k context length, it max out to 32k and using the following command we can increase it "export OLLAMA_CONTEXT_LENGTH=131072"
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#7279
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?