Smilyai-labs/Mira-1-XL

Full fine-tune vs Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled. Fine-tune: 100% of the weights changed, across 64 of 64 layers.

Tensors changed: lm_head 1/1, Embeddings 5/5, MLP 306/307, Attention 557/560, other 1/1, Norms 325/325

2 derivatives checked