qing-yao/ppt-pythia-1b-appendix-baseline-seed3407-stage2
Full fine-tune vs EleutherAI/pythia-1b. Fine-tune: 100% of the weights changed, across 16 of 16 layers.
Tensors changed: Embeddings 2/2, Norms 66/66, Attention 64/64, MLP 64/64
Full fine-tune vs EleutherAI/pythia-1b. Fine-tune: 100% of the weights changed, across 16 of 16 layers.
Tensors changed: Embeddings 2/2, Norms 66/66, Attention 64/64, MLP 64/64