mradermacher/Qwen3-72B-Instruct-GGUF
Full fine-tune vs QuixiAI/Qwen3-72B-Synthesis. mradermacher/Qwen3-72B-Instruct-GGUF is a GGUF (Q4_K_M) of different weights. Fine-tune: 100% of the weights changed, across 80 of 80 layers.
Tensors changed: Norms 257/321, MLP 240/240, Attention 320/320