Describe the Issue
-
I have set 1024 context size in launcher, I see 23744 in Lite web interface with max 8192 in slider. Max Output is 2048 whereas in log I see Generating (xxx / 820 tokens), so actual max size is 820.
-
Changing context size via Lite does not seem to attect actual context used by backend - after increase in Lite I see in log same CtxLimit, after decrease in Lite I don't see any more free memory. What is the point of changing Context Size in Lite? (changing Max Output does change results of backend).
Could Lite and backend be more 'synchronized' for Context Size?
TIA
Additional Information:
v1.111.2
P.S.
Do such issues belong more in https://github.com/LostRuins/lite.koboldai.net/issues?
Describe the Issue
I have set 1024 context size in launcher, I see 23744 in Lite web interface with max 8192 in slider. Max Output is 2048 whereas in log I see
Generating (xxx / 820 tokens), so actual max size is 820.Changing context size via Lite does not seem to attect actual context used by backend - after increase in Lite I see in log same CtxLimit, after decrease in Lite I don't see any more free memory. What is the point of changing Context Size in Lite? (changing Max Output does change results of backend).
Could Lite and backend be more 'synchronized' for Context Size?
TIA
Additional Information:
v1.111.2
P.S.
Do such issues belong more in https://github.com/LostRuins/lite.koboldai.net/issues?