Closed eliranwong closed 3 months ago
Hi there, you can set it to a large value (e.g. 16k, 1M) and Ollama will automatically use the largest context window the model was trained against. Hope this helps
Your suggestion does not work in some models I tested. It just hangs forever without response, if I set num_ctx too high.
Can I set num_ctx=-1 to use max possible context window?
so I don't need to change its value for different models