Universal Latent Space Alignment Across LLM Families via Linear Projection

We empirically confirm the Platonic Representation Hypothesis by demonstrating that hidden state representations from LLMs of entirely different families (Mistral-7B, Falcon-7B, Gemma-2-9B) align to Qwen2.5-7B's latent space with cosine similarity exceeding 0.99 using a single linear projection of approximately 4MB.

Keywords

latent space, linear projection, LLM, representation learning, model alignment, platonic hypothesis

Impact byBIP!

	selected citations These citations are derived from selected sources. This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).	0
	popularity This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.	Average
	influence This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).	Average
	impulse This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.	Average

Found an issue? Give us feedback

0

Average