Nexesenex/Llama_3.x_70b_SmarTracks_V1.01

Full fine-tune vs NexesMess/Llama_3.x_70b_SmarTracks_0.21. Fine-tune: 100% of the weights changed, across 80 of 80 layers.

Tensors changed: Norms 160/161, MLP 240/240, Attention 320/320

2 derivatives checked