mradermacher/7B-Curr-ReFT-GGUF
Same weights, converted (fewer layers) vs ZTE-AIM/7B-Curr-ReFT. mradermacher/7B-Curr-ReFT-GGUF matches part of ZTE-AIM/7B-Curr-ReFT (339 of its 729 tensors; the rest of ZTE-AIM/7B-Curr-ReFT has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/F32).
Tensors changed: Embeddings 0/1, Norms 0/57, MLP 0/84, Attention 0/196, lm_head 0/1