usermma/Holo-3.1-4B-abliterated-rdo
Partial fine-tune (fewer layers) vs Hcompany/Holo-3.1-4B. usermma/Holo-3.1-4B-abliterated-rdo matches part of Hcompany/Holo-3.1-4B (427 of its 724 tensors; the rest of Hcompany/Holo-3.1-4B has no counterpart). On the shared part: Partial change: 32% of the weights changed (attention, lm_head, mlp), in 9 of 32 layers.
Tensors changed: lm_head 1/1, Embeddings 0/1, MLP 27/96, Norms 0/105, Attention 24/224
2 derivatives checked
- mradermacher/Holo-3.1-4B-abliterated-rdo-i1-GGUF: Quantization
- mradermacher/Holo-3.1-4B-abliterated-rdo-GGUF: Same weights, converted