3MPER0RR/TinyLlama-3MPER0RR-abliterated
Abliterated vs TinyLlama/TinyLlama-1.1B-Chat-v1.0. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 15 of 22 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/45, MLP 15/66, Attention 11/88
4 derivatives checked
- mradermacher/TinyLlama-3MPER0RR-abliterated-i1-GGUF: Quantization
- mradermacher/TinyLlama-3MPER0RR-abliterated-GGUF: Same weights, converted
- mradermacher/TinyLlama-3MPER0RR-obliterated-GGUF: Same weights, converted
- 3MPER0RR/TinyLlama-3MPER0RR-abliterated-Q8_0-GGUF: Quantization