francoisfleuret

"there is no understanding, only prediction"

I have no idea what this means.

This is 300M parameters model (1/1300th of the big llama-3) trained with 5M frames with 12 days of a GTX4090.

This is what a big tech company was doing in 2015.

The same stuff at industrial scale à la large LLMs would be absolutely mind blowing.

This is a minor update of the "Little Book of Deep Learning", a phone-formatted free book for people with a STEM background that happened to be quite popular (580k+ download in a year).

This update adds in particular a chapter about methods to run or fine-tune LLMs with limited compute and memory (prompt engineering, quantization, low-rank adapters, and model merging).