fr0stbit3/laya-multilingual-gguf
Same weights, converted (fewer layers) vs convaiinnovations/laya-multilingual. fr0stbit3/laya-multilingual-gguf matches part of convaiinnovations/laya-multilingual (36 of its 170 tensors; the rest of convaiinnovations/laya-multilingual 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