ykarout/Mixtral-8x7B-DeepSeek-R1-Distill
LoRA-style vs mistralai/Mixtral-8x7B-Instruct-v0.1. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, o_proj, q_proj, v_proj, w1, w2, w3 changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, MLP 768/768, other 0/32, Norms 0/65, Attention 128/128, lm_head 0/1
2 derivatives checked
- mradermacher/Mixtral-8x7B-DeepSeek-R1-Distill-i1-GGUF: Quantization
- mradermacher/Mixtral-8x7B-DeepSeek-R1-Distill-GGUF: Quantization