NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO
Full fine-tune vs mistralai/Mixtral-8x7B-v0.1. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: MLP 768/768, other 32/32, Norms 37/65, Attention 128/128
10 derivatives checked
- TheBloke/Nous-Hermes-2-Mixtral-8x7B-DPO-GGUF: Not comparable yet
- mradermacher/Nous-Hermes-2-Mixtral-8x7B-DPO-i1-GGUF: Quantization
- mradermacher/Nous-Hermes-2-Mixtral-8x7B-DPO-GGUF: Quantization
- second-state/Nous-Hermes-2-Mixtral-8x7B-DPO-GGUF: Not comparable yet
- TheBloke/Nous-Hermes-2-Mixtral-8x7B-DPO-GPTQ: Quantization
- Saxo/Linkbricks-Horizon-AI-Korean-Mixtral-sft-dpo-8x7b: Full fine-tune
- TheBloke/Nous-Hermes-2-Mixtral-8x7B-DPO-AWQ: Quantization
- Saxo/Linkbricks-Horizon-AI-Korean-Advanced-56B: Full fine-tune
- rhplus0831/maid-yuzu-v3: Full fine-tune
- rhplus0831/maid-yuzu-v3-alter: Full fine-tune
2 other repos hold byte-for-byte identical weight files.