3 ms·
You can enable the 1 million token context window and adjust when it compacts in your config. > model_context_window = 1000000 > model_auto_compact_token_limi
by swingboy 29d ago
You can enable the 1 million token context window and adjust when it compacts in your config.
> model_context_window = 1000000
> model_auto_compact_token_limit = 900000
I believe it does consume your usage a bit faster though.
- fnordpiglet 29d agoUnless things changed this used to work but was blocked. You also used to be able to force in a model catalog to get around the token cap. Each, as of at least April, were closed off and ineffectual. I’ve not tried lately so you may be right.