nvidia/llama-3.1-nemoguard-8b-content-safety
Adapter vs meta-llama/Llama-3.1-8B-Instruct. nvidia/llama-3.1-nemoguard-8b-content-safety is a LoRA adapter, not full weights: rank 16, on down_proj, gate_proj, k_proj, lm_head, o_proj, q_proj, up_proj, v_proj in 32 layers (225 adapted matrices). It modifies meta-llama/Llama-3.1-8B-Instruct at load time.
1 other repos hold byte-for-byte identical weight files.