jukofyork/Dark-Miqu-70B
Partial fine-tune vs 152334H/miqu-1-70b-sf. Partial change: 85% of the weights changed (attention, lm_head, mlp, norms), in 80 of 80 layers.
Tensors changed: Embeddings 0/1, Norms 128/161, MLP 207/240, Attention 272/320, lm_head 1/1
4 derivatives checked
- backyardai/Dark-Miqu-70B-GGUF: Same weights, converted
- mradermacher/Dark-Miqu-70B-i1-GGUF: Quantization
- mradermacher/Dark-Miqu-70B-GGUF: Quantization
- softwareweaver/Twilight-Miqu-146B: Full fine-tune (fewer layers)