3 ms·
So I can feed a lossy text to an LLM and it can “delossy” it? Does the lossy text take up less space so this could work as a weird compression?
by maCDzP 3y ago
So I can feed a lossy text to an LLM and it can “delossy” it?
Does the lossy text take up less space so this could work as a weird compression?
- datascienced 3y agoUse an embedding, then quantize (i.e. rounding) maybe?