kurakurai/Luth-1.7B-Instruct
Full fine-tune vs Qwen/Qwen3-1.7B. Fine-tune: 97% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 105/113, MLP 81/84, Attention 108/112
5 derivatives checked
- prithivMLmods/Luth-Instruct-GGUF: Not comparable yet
- mradermacher/Luth-1.7B-Instruct-i1-GGUF: Quantization
- DevQuasar/kurakurai.Luth-1.7B-Instruct-GGUF: Same weights, converted
- mradermacher/Luth-1.7B-Instruct-GGUF: Same weights, converted
- Ronaldodev/finanfa-ai-luth-1.7b: LoRA-style