Sub-1-Bit LLM Compression via Latent Factorization

Perf goes from 80% to 47% on Wikitext-2. Also no comparisons to FP4 solutions that are able to maintain or exceed perf on the same dataset 80% perf with a 4.25-4.5 big budget.

I think more meaningful thing here would be a hybrid solution that went down to sub-bit representations when the informational representation does not need it (for example later layers) that still maintains task performance

Can this produce a useful model? So far 1 bit quants have been less useful than smaller models that use the same memory

Their paper shows this comes with huge quality loss, but that doesn't make it a negative result by any means

Has anybody tested this? Are there any available computed models to test?

Thought this was going to be on the original Little Bit paper, always nice to find out about a surprise sequel!

Has anyone tried this on apple silicon M1-5? Any benchmarks/comps?