KIEFERSA/Sophea-Nemo-Embedding

Full fine-tune vs nvidia/Nemotron-3-Embed-1B-BF16. Fine-tune: 100% of the weights changed, across 16 of 16 layers.

Tensors changed: Embeddings 1/1, Norms 1/33, MLP 48/48, Attention 64/64

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse