Reset token budget after every user intervention.

In interactive mode, every time the model has to respond to user input
it has an increasingly reduced token budget, eventually generating only
a few words before stopping. The token budget in interactive should
apply to every batch of tokens after user intervention, not globally

main.cpp

@@ -1054,12 +1054,12 @@ int main(int argc, char ** argv) {
                     embd_inp.insert(embd_inp.end(), inp_sfx.begin(), inp_sfx.end());
                 }
 
-                remaining_tokens -= line_inp.size();
+                remaining_tokens = params.n_predict - line_inp.size();
 
                 input_noecho = true; // do not echo this again
-            }
                 is_interacting = false;
             }
+        }
 
         // end of text token
         if (embd.back() == EOS_TOKEN_ID) {