rinna/bilingual-gpt-neox-4b-8k
Full fine-tune vs rinna/bilingual-gpt-neox-4b. Fine-tune: 98% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 2/2, Norms 146/146, Attention 144/252, MLP 144/144
Full fine-tune vs rinna/bilingual-gpt-neox-4b. Fine-tune: 98% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 2/2, Norms 146/146, Attention 144/252, MLP 144/144