LiquidAI/LFM2-1.2B config.json

LFM2 config exposes full_attn_idxs [2,5,8,10,12,14] plus conv_L_cache 3 and conv_dim 2048: 10 short conv layers and 6 full-attention layers. This is the hybrid architecture. Primary source for every architecture field in this record; values read from the repo's config.json, not from a model card summary. W10 LOWCONFIDENCE PROVENANCE AUDIT 2026-10-10: re-classified UNCHECKED after an actual re-fetch. Fetched over HTTPS from this host: HTTP 200, the returned file is live JSON and parses. Every architecture field this note names was then read back out of the live config.json and matches the note's value field-for-field. The 0.5 rating therefore marked a lookup nobody had run, not a property of the source: it is the model publisher's own repository, so it is a primary-shaped source, and the one caveat is that it is a mirror for the license-gated originals (see 'a-hosting-platform-is-not-a-vendor' and the gated-repo records). Raised 0.5 -> 0.88, status draft -> verified.