qihoo360/Light-IF-32B

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 206/257, MLP 192/192, Attention 256/256, lm_head 1/1

2 derivatives checked

• mradermacher/Light-IF-32B-i1-GGUF: Quantization
• mradermacher/Light-IF-32B-GGUF: Quantization

On the Hub · Downloads on Model Pulse