OccultAI/Trixster-MoE-14B-A8B-v1
Full fine-tune vs Babsie/CrossroadsLoki-MoE-2x8B. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: MLP 192/192, Norms 65/65, Attention 128/128, other 32/32
2 derivatives checked
- mradermacher/Trixster-MoE-14B-A8B-v1-i1-GGUF: Quantization
- mradermacher/Trixster-MoE-14B-A8B-v1-GGUF: Quantization