rubenroy/Gilgamesh-72B
Full fine-tune vs Qwen/Qwen2.5-72B-Instruct. Fine-tune: 98% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 0/1, Norms 116/161, MLP 240/240, Attention 558/560, lm_head 1/1
3 derivatives checked
- bartowski/rubenroy_Gilgamesh-72B-GGUF: Quantization
- mradermacher/Gilgamesh-72B-i1-GGUF: Quantization
- mradermacher/Gilgamesh-72B-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.