# @timfduffy — 2025-08-29

♥0 ↻0 · https://x.com/timfduffy/status/1961576898947158082

@tessera_antra The residual stream certainly provides coherence between layers of a single forward pass, but it is discarded at the end of a forward pass, so the asemantic information between tokens is just in the KV cache. I don't think this undermines the overall argument though.

tags: author:timfduffy, kind:tweet, thread-context, year:2025