I wonder how consistent is the decompression and how much information is lost in the process.
i’d guess they could hyper optimise for “perceived difference” rather than data loss specifically… they do a pretty good job of generating something from nothing, so i’d say with enough data they’d probably generate a pretty reasonable facsimile of “standard” stuff
An LLM can’t know what difference a person has perceived.
Honestly? I’m not super surprised by this. The human brain (and I assume brains in general) are really good at data compression. Considering neural networks are more or less meant to mimic different aspects of the human brain, it doesn’t surprise me too much that they’d be really good at data compression as well.
So like, mp3, gzip and zstd? Why would you use a LLM for compression??
The research specifically looked at lossless algorithms, so gzip
“For example, the 70-billion parameter Chinchilla model impressively compressed data to 8.3% of its original size, significantly outperforming gzip and LZMA2, which managed 32.3% and 23% respectively.”
However they do say that it’s not especially practical at the moment, given that gzip is a tiny executable compared to the many gigabytes of the LLM’s dataset.
Do you need the dataset to do the compression? Is the trained model not effective on its own?
Well from the article a dataset is required, but not always the heavier one.
Tho it doesn’t solve the speed issue, where the llm will take a lot more time to do the compression.
gzip can compress 1GB of text in less than a minute on a CPU, an LLM with 3.2 million parameters requires an hour to compress
deleted by creator
Gavin Belson has entered the chat