CultriX/MoNeuTrix-7B-v1
Full fine-tune vs Kukedlc/NeuralMaxime-7B-slerp. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: MLP 96/96, Attention 128/128, Norms 65/65, Embeddings 1/1, lm_head 1/1
4 derivatives checked
- CultriX/NeuralMona_MoE-4x7B: MoE built from dense models
- mradermacher/MoNeuTrix-7B-v1-i1-GGUF: Quantization
- mradermacher/MoNeuTrix-7B-v1-GGUF: Same weights, converted
- CultriX/MoNeuTrix-MoE-4x7B: MoE built from dense models