by jbay808 on 6/13/23, 3:40 PM with 20 comments
by HarHarVeryFunny on 6/15/23, 2:18 PM
I understand how transformers work, but my mental model is that a transformer is the processor and the LLM is an application that runs on it. After all, transformers can be trained to do lots of things, and what it learns when trained with a "predict next word" LLM objective is going to differ from what it learns (and hence operates?) in a different setting.
There have been various LLM interpretation papers analyzing aspects of them, such as the discovery of pairs of consecutive layer attention heads acting as "search and copy" "induction heads", and analysis of the linear layers as key-value stores, which perhaps leads to another weak abstraction of the linear layers as storing knowledge and perhaps the reasoning "program", with the "attention" layers being the mechanism being programmed to do the data tagging/shuffling ?
No doubt there's a lot more to be discovered about how these LLMs are operating - perhaps a wider variety of primitives built out of attention heads other than just induction heads ? It seems a bit early to be building a high level model of the primitives these LLMs have learnt, and not sure if attempting a crude transformer-level model really works given how the residual context is additive - it's not just tokens being moved around.
by inciampati on 6/15/23, 10:44 AM
by rolisz on 6/15/23, 10:36 AM
There is an interpreter for a RASP like language if you want to try it out: https://srush.github.io/raspy/
And deepmind published a compiler from RASP to Transformer weights: https://github.com/deepmind/tracr
by ljlolel on 6/15/23, 10:39 AM