gaianet/moxin-chat-7b-GGUF

Same weights, converted (fewer layers) vs moxin-org/Moxin-7B-Chat. moxin-org/Moxin-7B-Chat matches part of gaianet/moxin-chat-7b-GGUF (16 of its 327 tensors; the rest of gaianet/moxin-chat-7b-GGUF has no counterpart). On the shared part: Same weights stored in another dtype (F16 → F16/F32), plus 311 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/2, MLP 0/5, Attention 0/8

On the Hub · Downloads on Model Pulse