Ollama local models context token length always displays 4096 even when LLMs support more

[mysterium-coniunctionis]

For some reason, local Ollama models that support contexts larger than 4096 tokens do not seem to adjust to the greater token length on Chatbot-UI. Has this been addressed anywhere? If yes, will you point me in the right direction to fix this on my localhost instance?

[mckaywrigley]

Do you know if the Ollama API gives us model context length? If so I can fix this in like 5min.

[mysterium-coniunctionis]

There are a few that vary as far as I can tell at various levels - as low as 2K and as high as 128K. I found a couple that were easy to search for and I'll add more to this thread as I have a bit of time to dig through the docs @ https://ollama.ai/library.

I'm not sure if this is what you mean, but here are a few:

yarn-mistral & yarn-llama2 both come in 64K & 128K versions

vicuna comes in 2K-16K versions

everythinglm comes a 16K version

alfred comes in an 8K version