I think the max token problem is because we count max-tokens after we restricted the exploration... This is obviously a problem...
Counting the first state for queries that are satisfies initially is probably just adding +1 a few places... Nothing big...
I'll try to take a look at this...
I think the max token problem is because we count max-tokens after we restricted the exploration... This is obviously a problem...
Counting the first state for queries that are satisfies initially is probably just adding +1 a few places... Nothing big...
I'll try to take a look at this...