KaraKaraWarehouse/spiral-da-HYAH-Qwen2.5-72b

Full fine-tune vs AXCXEPT/EZO-Qwen2.5-72B-Instruct. Fine-tune: 100% of the weights changed, across 80 of 80 layers.

Tensors changed: Embeddings 1/1, Norms 51/161, MLP 240/240, Attention 502/560, lm_head 1/1

1 derivatives checked

• mradermacher/spiral-da-HYAH-Qwen2.5-72b-i1-GGUF: Quantization

On the Hub · Downloads on Model Pulse