infly/INFIndo-Qwen3-32B-Preview

Full fine-tune vs Qwen/Qwen3-32B. Fine-tune: 98% of the weights changed, across 64 of 64 layers.

Tensors changed: Embeddings 0/1, Norms 149/257, MLP 192/192, Attention 256/256, lm_head 1/1

2 derivatives checked