4 ms·
A few months ago I asked why semantic representation rather than text wasn't used, since natural language seems quite a lossy representation for semantic concep
by gavinray 8d ago
A few months ago I asked why semantic representation rather than text wasn't used, since natural language seems quite a lossy representation for semantic concepts:
https://news.ycombinator.com/item?id=47195212 https://news.ycombinator.com/item?id=47195212
I wouldn't have thought to use it for LLM-to-LLM communication, though
- msdz 8d agoMy guess is that if you go with something other than readable text as the “thought layer”, observability becomes impossible. Which is not necessarily something the humans training a model would want re/ alignment.
- TeMPOraL 7d agoIndeed this is one of the key problems here. If models start doing CoT in "neuralese", or use it when talking with each other, we lose what little observability we have.