eduardo-alvarez/GemmaSheep-2B-LORA-TUNED
Adapter vs google/gemma-2b. eduardo-alvarez/GemmaSheep-2B-LORA-TUNED is a LoRA adapter, not full weights: rank 32, on down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj in 18 layers (126 adapted matrices). It modifies google/gemma-2b at load time.