Open-Bee/Bee-8B-SFT
Full fine-tune (fewer layers) vs Qwen/Qwen3-8B. Qwen/Qwen3-8B matches part of Open-Bee/Bee-8B-SFT (397 of its 827 tensors; the rest of Open-Bee/Bee-8B-SFT has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Norms 145/145, MLP 108/108, Attention 144/144