Is your feature request related to a problem? Please describe.
Not really a problem, but rather an optimisation. I'm currently hosting Llama3 and would like to reuse this model for every task like User queries, Chat title generation, Web search. The issue I have is that I have set larger context length for the model in the Workspaces, which differs to the one (default) used in Task Model. The result is that Ollama constantly reloads the model, since the LLM parameters are different. If they are the same, the model will be kept in the GPU memory and will be much efficient.

Describe the solution you'd like
A best solultion would be to have an option to use the workspace model's parameters in Set Task Model, or less ideal to have a way to override the default LLM parameters using environment variables.

Describe alternatives you've considered
The only alternative I have found is to use very small model that can fit in the GPU memory along with the larger and use it as Task Model.