allenai/llama-3-tulu-2-70b-uf-mean-rm
Full fine-tune vs allenai/llama-3-tulu-2-70b. Fine-tune: 98% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 0/1, Norms 136/161, MLP 240/240, Attention 320/320
Full fine-tune vs allenai/llama-3-tulu-2-70b. Fine-tune: 98% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 0/1, Norms 136/161, MLP 240/240, Attention 320/320