nvidia/Nemotron-H-47B-Base-8K
Full fine-tune vs nvidia/Nemotron-H-56B-Base-8K. Fine-tune: 100% of the weights changed, across 98 of 98 layers.
Tensors changed: Embeddings 1/1, other 189/189, Norms 126/126, lm_head 1/1
Full fine-tune vs nvidia/Nemotron-H-56B-Base-8K. Fine-tune: 100% of the weights changed, across 98 of 98 layers.
Tensors changed: Embeddings 1/1, other 189/189, Norms 126/126, lm_head 1/1