Jeesup/svdsafety_l2_addbase_top256
Full fine-tune vs meta-llama/Llama-2-7b-chat-hf. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/1, Norms 65/65, MLP 96/96, Attention 128/128, lm_head 1/1
Full fine-tune vs meta-llama/Llama-2-7b-chat-hf. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/1, Norms 65/65, MLP 96/96, Attention 128/128, lm_head 1/1