geohot / twitchctw
Licence: other
compression = AI
Stars: ✭ 46
CTW compressor. How long before we beat gzip gzip: 3728 bytes <-- not easy to beat actually xz: 3656 bytes ## comparing to existing CTW implementations # https://github.com/cberzan/ctw.git ctw_baseline This is the authors imp? gets 3484, so CTW can do better than gzip Though this code has tons of hacks, notably MAXCOUNTS If you remove MAXCOUNTS it's crap, but this could be other things That's the annoying thing about this problem, little bug = slightly worse And I don't understand it's tree depth # https://github.com/fumin/ctw.git gets 4616, err, not good # https://github.com/gkassel/pyaixi CTW implementation seems numerically unstable... # https://github.com/mgbellemare/SkipCTS/tree/master/python logprob of data is 2988 bytes with 16 context, haven't tried in compressor yet Maybe this isn't as easy as I thought... Update === It is a working compressor (run ./test.sh) but it's performance is quite bad. Maybe it's bugs? One day we'll work through the proofs in the paper.
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].