EleutherAI/deep-ignorance-e2e-strong-filter
Full fine-tune vs EleutherAI/deep-ignorance-pretraining-stage-unfiltered. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 2/2, Attention 128/128, Norms 130/130, MLP 128/128
1 other repos hold byte-for-byte identical weight files.