computer networks – Can you explain this token bucket example?

Suppose the token-bucket specification is TB(1/3 packet/ms, 4 packets), and packets arrive at the following times, with the bucket initially full:

0, 0, 0, 2, 3, 6, 9, 12
After all the T=0 packets are processed, the bucket holds 1 token. By the time the fourth packet arrives at T=2, the bucket volume has risen to 1 2/3; it immediately drops to 2/3 when packet 4 is sent. By T=3, the bucket volume has reached 1 and the fifth packet can be sent. The bucket is now empty, but fortunately the remaining packets arrive at 3-ms intervals and can all be sent.

Source-:
http://intronetworks.cs.luc.edu/current/html/tokenbucket.html

What I know-:

Token bucket algorithm-:
1 token added to bucket in every $Delta t$ time.

For a packet to be transmitted, it must destroy the token.
That is all I know about token bucket algorithm

As far as the question is considered-:

TB(r,$B_{max};)$

Where

r=token fill rate token/sec

$B_{max};$=Bucket Capacity