allenai/Bolmo-1B
Full fine-tune vs allenai/Bolmo-1B-Stage1. Fine-tune: 100% of the weights changed, across 16 of 16 layers.
Tensors changed: MLP 63/63, Norms 82/82, Attention 66/66, other 49/49, Embeddings 1/2, lm_head 1/1
Full fine-tune vs allenai/Bolmo-1B-Stage1. Fine-tune: 100% of the weights changed, across 16 of 16 layers.
Tensors changed: MLP 63/63, Norms 82/82, Attention 66/66, other 49/49, Embeddings 1/2, lm_head 1/1