OBLITERATUS/Ornith-1.5-9B-OBLITERATED
Partial fine-tune vs ornith-ai/Ornith-1.5-9B. Partial change: 51% of the weights changed (attention, lm_head, mlp), in 19 of 32 layers.
Tensors changed: Embeddings 0/5, Norms 0/221, Attention 48/336, MLP 57/211, lm_head 1/1, other 0/1
6 derivatives checked
- mradermacher/Ornith-1.5-9B-OBLITERATED-i1-GGUF: Quantization
- mradermacher/Ornith-1.5-9B-OBLITERATED-GGUF: Same weights, converted
- DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED-GGUF: LoRA-style
- DuoNeural/Cyber-Ornith-1.5-9B-OBLITERATED: LoRA-style (fewer layers)
- 0xSojalSec/Uncensored-Cyber-Ornith-1.5-9B: LoRA-style (fewer layers)
- Mantisec/Ornith-1.5-9B-OBLITERATED-FP16: Same weights, converted
3 other repos hold byte-for-byte identical weight files.