bobber/routangseng-0.8b-hottake
Full fine-tune (fewer layers) vs huihui-ai/Huihui-Qwen3.5-0.8B-abliterated. bobber/routangseng-0.8b-hottake matches part of huihui-ai/Huihui-Qwen3.5-0.8B-abliterated (320 of its 488 tensors; the rest of huihui-ai/Huihui-Qwen3.5-0.8B-abliterated has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Attention 134/168, Norms 27/79, Embeddings 1/1, MLP 72/72
1 derivatives checked
- mradermacher/routangseng-0.8b-hottake-GGUF: Same weights, converted