Ollama setup for Cline
Sometimes Ollama cannot be used by cline, because the size of model context is too small, and cline will complain about it.
The size of model context can be increased by altering the num_ctx parameter. In the following steps i’ll create a new model from an existing one, with increased model context.
First we need to export the configuration of the model for which we want to increase the context size:
ollama show --modelfile "deepseek-r1:14b" > deepseek-r1-14b-32k.ModelfileOpen the created file deepseek-r1-14b-32k.Modelfile using a text editor and the change the FROM line
original content:
FROM ollama/llama2:latest # or other default content
New content:
FROM deepseek-r1:14b
Manually add the parameters, by appending to the end of the configuration file parameter value
PARAMETER num_ctx 32768
Create new model from the custom config file
ollama create deepseek-r1-14b-32k -f deepseek-r1-14b-32k.Modelfile
Chosse in Cline setting the newly generated model, and it will not complain about the context size.