@Bonk:

As always, you mods from GTru deny any mistakes. If every user tells you that they only have problems uploading/seeding on your tracker, doesn't that ring a bell?

The questioner is asking for the forum community to provide him with a knowable statement about his problem (i.e., "I know what caused your problem.") It has been said that a well-asked question is already half-answered. The forum's failure to provide this knowable statement is a failure shared with the questioner himself, as key information about the problem was omitted from his query. His problem was thus presented as the following:

@shiftless:

When I'm downloading a file, I'm able to upload at full speed just fine. After the download has completed the upload rate crashes to no activity at all most of the time.

He has presented a scenario whereby uploading rates mysteriously change during the uploading of certain torrents. Missing here are key details about the torrent being uploaded when the occurance was noticed by him.

The age of the torrent at the moment he entered the swarm was a detail not provided. Since any gainful return on seeding depends on bandwidth and torrent age, upload rates tend to be significantly higher when the user arrives early to a torrent, and tend to increase linearly with bandwidth over a given period. What was the duration of this particular event? Was this noticeable drop in upload speed an event with a short duration of time, or was it graduated, occurring over a wider and more cumulative breadth of time?

If these instances of "drop off" were noticed on only a few occasions, should that, in itself, be suspect? People are prone to see patterns where there are none (such as in occurrences of "upload speed drop off") and the mind may attempt to attribute them to some sort of correlation (claiming it's a problem with the client, the tracker, or the ISP) when in fact it may be part and parcel of the normal bandwidth variability inherent in all torrenting.

This "pattern bias" aside, the most important thing to realize about the questioner's issue is that the torrent's age should always factor in any discussion of upload bandwidth (speed). Being able to determine the age of the uploading torrent at the moment the drop off occurred may likely have been the critical component to providing this questioner with the knowable statement he sought. Since we were not provided with it, the only knowable aspect of this questioner's problem is that we will never know what caused it.