Frowning/Mag-Mell-R1-21B
Partial fine-tune (fewer layers) vs inflatebot/MN-12B-Mag-Mell-R1. inflatebot/MN-12B-Mag-Mell-R1 matches part of Frowning/Mag-Mell-R1-21B (363 of its 633 tensors; the rest of Frowning/Mag-Mell-R1-21B has no counterpart). On the shared part: Partial change: 22% of the weights changed (attention, mlp, norms), in 10 of 40 layers.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 20/81, MLP 30/120, Attention 40/160
2 derivatives checked
- mradermacher/Mag-Mell-R1-21B-i1-GGUF: Quantization
- mradermacher/Mag-Mell-R1-21B-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.