Vikhrmodels/Vikhr-YandexGPT-5-Lite-8B-it
Abliterated vs yandex/YandexGPT-5-Lite-8B-pretrain. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 32 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/65, MLP 32/96, Attention 32/128, lm_head 0/1
3 derivatives checked
- MaziyarPanahi/Vikhr-YandexGPT-5-Lite-8B-it-GGUF: Same weights, converted
- DevQuasar-9/Vikhrmodels.Vikhr-YandexGPT-5-Lite-8B-it-GGUF: Same weights, converted
- tensorblock/Vikhrmodels_Vikhr-YandexGPT-5-Lite-8B-it-GGUF: Quantization