sz14/cRia-LM-75M-Instruct
Full fine-tune vs sz14/cRia-LM-75M. Fine-tune: 100% of the weights changed, across 13 of 13 layers.
Tensors changed: Attention 166/166, Norms 27/27, MLP 105/105, Embeddings 2/2, other 1/1
Full fine-tune vs sz14/cRia-LM-75M. Fine-tune: 100% of the weights changed, across 13 of 13 layers.
Tensors changed: Attention 166/166, Norms 27/27, MLP 105/105, Embeddings 2/2, other 1/1