fr0stbit3/laya-gguf
Same weights, converted (fewer layers) vs convaiinnovations/laya. fr0stbit3/laya-gguf matches part of convaiinnovations/laya (36 of its 206 tensors; the rest of convaiinnovations/laya has no counterpart). On the shared part: Same weights stored in another dtype (F16/F32 → F32).
Tensors changed: other 0/20, Norms 0/8, Attention 0/8