kurakurai/Luth-0.6B-Instruct
Full fine-tune vs Qwen/Qwen3-0.6B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 104/113, MLP 84/84, Attention 112/112
3 derivatives checked
- mradermacher/Luth-0.6B-Instruct-i1-GGUF: Quantization
- DevQuasar/kurakurai.Luth-0.6B-Instruct-GGUF: Same weights, converted
- mradermacher/Luth-0.6B-Instruct-GGUF: Same weights, converted