Elfrino/BinaryBlush-33B
Full fine-tune (fewer layers) vs Elfrino/XwinXtended-20B. Elfrino/XwinXtended-20B matches part of Elfrino/BinaryBlush-33B (559 of its 939 tensors; the rest of Elfrino/BinaryBlush-33B has no counterpart). On the shared part: Fine-tune: 99% of the weights changed, across 62 of 62 layers.
Tensors changed: Norms 123/125, MLP 184/186, Attention 244/248
1 derivatives checked
- mradermacher/BinaryBlush-33B-i1-GGUF: Quantization