dphn/dolphin-2.9.1-mixtral-1x22b
Full fine-tune vs mistral-community/Mixtral-8x22B-v0.1. Fine-tune: 100% of the weights changed, across 56 of 56 layers.
Tensors changed: MLP 168/168, Attention 224/224, Norms 100/113
4 derivatives checked
- mradermacher/dolphin-2.9.1-mixtral-1x22b-i1-GGUF: Quantization
- mradermacher/dolphin-2.9.1-mixtral-1x22b-GGUF: Quantization
- legraphista/dolphin-2.9.1-mixtral-1x22b-IMat-GGUF: Same weights, converted
- InferenceIllusionist/dolphin-2.9.1-mixtral-1x22b-iMat-GGUF: Quantization