TheDrummer/Tunguska-39B-v1

Full fine-tune vs SteelStorage/BA-Zephyria-39b. Fine-tune: 99% of the weights changed, across 99 of 99 layers.

Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 124/199, MLP 297/297, Attention 396/396

1 derivatives checked

• bartowski/Tunguska-39B-v1-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse