mnoukhov/pythia410m-sft-tldr
Full fine-tune vs EleutherAI/pythia-410m-deduped. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Embeddings 2/2, Norms 98/98, Attention 96/96, MLP 96/96
Full fine-tune vs EleutherAI/pythia-410m-deduped. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Embeddings 2/2, Norms 98/98, Attention 96/96, MLP 96/96