HuggingFaceTB/nanowhale-100m

Full fine-tune vs HuggingFaceTB/nanowhale-100m-base. Fine-tune: 100% of the weights changed, across 8 of 8 layers.

Tensors changed: lm_head 1/1, Embeddings 1/1, other 3/3, Attention 64/72, Norms 33/33, MLP 152/160

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

On the Hub · Downloads on Model Pulse