jsfs11/MoEv4Config-TIESwithRescale-7b
Full fine-tune vs Kukedlc/NeuTrixOmniBe-7B-model-remix. Fine-tune: 98% of the weights changed, across 32 of 32 layers.
Tensors changed: MLP 96/96, Norms 63/65, Attention 124/128, lm_head 1/1, Embeddings 0/1
2 derivatives checked
- mradermacher/MoEv4Config-TIESwithRescale-7b-i1-GGUF: Quantization
- mradermacher/MoEv4Config-TIESwithRescale-7b-GGUF: Same weights, converted