variants/cpu-int4/train/training_config.json

{"requires_grad": ["backbone.model.layers.0.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.0.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.0.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.0.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.1.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.1.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.1.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.1.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.2.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.2.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.2.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.2.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.3.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.3.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.3.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.3.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.4.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.4.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.4.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.4.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.5.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.5.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.5.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.5.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.6.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.6.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.6.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.6.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.7.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.7.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.7.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.7.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.8.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.8.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.8.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.8.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.9.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.9.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.9.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.9.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.10.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.10.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.10.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.10.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.11.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.11.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.11.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.11.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.12.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.12.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.12.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.12.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.13.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.13.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.13.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.13.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.14.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.14.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.14.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.14.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.15.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.15.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.15.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.15.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.16.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.16.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.16.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.16.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.17.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.17.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.17.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.17.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.18.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.18.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.18.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.18.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.19.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.19.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.19.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.19.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.20.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.20.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.20.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.20.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.21.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.21.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.21.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.21.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.22.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.22.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.22.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.22.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.23.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.23.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.23.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.23.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.24.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.24.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.24.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.24.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.25.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.25.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.25.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.25.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.26.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.26.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.26.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.26.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.27.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.27.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.27.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.27.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.28.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.28.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.28.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.28.self_attn.v_proj.lora_B.lora.weight", "backbone.model.layers.29.self_attn.q_proj.lora_A.lora.weight", "backbone.model.layers.29.self_attn.q_proj.lora_B.lora.weight", "backbone.model.layers.29.self_attn.v_proj.lora_A.lora.weight", "backbone.model.layers.29.self_attn.v_proj.lora_B.lora.weight"], "trainable_tensor_specs": {"backbone.model.layers.0.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.0.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.0.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.0.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.1.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.1.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.1.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.1.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.2.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.2.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.2.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.2.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.3.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.3.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.3.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.3.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.4.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.4.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.4.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.4.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.5.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.5.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.5.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.5.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.6.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.6.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.6.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.6.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.7.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.7.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.7.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.7.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.8.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.8.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.8.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.8.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.9.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.9.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.9.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.9.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.10.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.10.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.10.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.10.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.11.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.11.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.11.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.11.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.12.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.12.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.12.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.12.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.13.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.13.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.13.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.13.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.14.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.14.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.14.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.14.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.15.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.15.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.15.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.15.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.16.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.16.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.16.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.16.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.17.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.17.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.17.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.17.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.18.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.18.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.18.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.18.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.19.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.19.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.19.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.19.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.20.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.20.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.20.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.20.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.21.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.21.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.21.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.21.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.22.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.22.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.22.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.22.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.23.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.23.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.23.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.23.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.24.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.24.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.24.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.24.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.25.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.25.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.25.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.25.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.26.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.26.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.26.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.26.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.27.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.27.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.27.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.27.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.28.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.28.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.28.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.28.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}, "backbone.model.layers.29.self_attn.q_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.29.self_attn.q_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [576, 8]}, "backbone.model.layers.29.self_attn.v_proj.lora_A.lora.weight": {"dtype": "float32", "shape": [8, 576]}, "backbone.model.layers.29.self_attn.v_proj.lora_B.lora.weight": {"dtype": "float32", "shape": [192, 8]}}, "peft_mapping": {"base_model.model.model.layers.0.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.0.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.0.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.0.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.0.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.0.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.1.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.1.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.1.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.1.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.1.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.1.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.2.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.2.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.2.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.2.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.2.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.2.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.3.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.3.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.3.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.3.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.3.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.3.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.4.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.4.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.4.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.4.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.4.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.4.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.5.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.5.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.5.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.5.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.5.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.5.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.6.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.6.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.6.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.6.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.6.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.6.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.7.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.7.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.7.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.7.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.7.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.7.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.8.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.8.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.8.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.8.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.8.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.8.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.9.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.9.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.9.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.9.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.9.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.9.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.10.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.10.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.10.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.10.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.10.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.10.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.11.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.11.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.11.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.11.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.11.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.11.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.12.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.12.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.12.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.12.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.12.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.12.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.13.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.13.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.13.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.13.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.13.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.13.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.14.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.14.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.14.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.14.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.14.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.14.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.15.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.15.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.15.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.15.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.15.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.15.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.16.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.16.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.16.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.16.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.16.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.16.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.17.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.17.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.17.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.17.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.17.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.17.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.18.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.18.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.18.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.18.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.18.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.18.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.19.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.19.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.19.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.19.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.19.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.19.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.20.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.20.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.20.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.20.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.20.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.20.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.21.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.21.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.21.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.21.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.21.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.21.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.22.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.22.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.22.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.22.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.22.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.22.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.23.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.23.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.23.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.23.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.23.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.23.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.24.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.24.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.24.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.24.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.24.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.24.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.25.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.25.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.25.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.25.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.25.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.25.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.26.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.26.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.26.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.26.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.26.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.26.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.27.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.27.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.27.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.27.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.27.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.27.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.28.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.28.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.28.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.28.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.28.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.28.self_attn.v_proj.lora_B.lora"}, "base_model.model.model.layers.29.self_attn.q_proj": {"adapter_A": "base_model.model.model.layers.29.self_attn.q_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.29.self_attn.q_proj.lora_B.lora"}, "base_model.model.model.layers.29.self_attn.v_proj": {"adapter_A": "base_model.model.model.layers.29.self_attn.v_proj.lora_A.lora", "adapter_B": "base_model.model.model.layers.29.self_attn.v_proj.lora_B.lora"}}, "rank": 8, "alpha": 8, "peft_target": ["q_proj", "v_proj"], "trainable_parameter_count": 460800, "source_parameter_count": 134515008}