ohmysimo/Wren

Partial fine-tune vs ukisai/Swift1.5-Qwen3.8-Flash-Next. Partial change: 0% of the weights changed (mlp, norms), in 48 of 48 layers.

Tensors changed: Norms 125/294, other 0/5, Attention 0/600, MLP 48/448, Embeddings 0/135, lm_head 0/1

1 derivatives checked