{ "measurement": { "model.layers.0": { "accuracy": 0.9724436340038665, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.1": { "accuracy": 0.9742336794734001, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.2": { "accuracy": 0.9743453479022719, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.3": { "accuracy": 0.9728944995440543, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.4": { "accuracy": 0.9719419548637234, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.5": { "accuracy": 0.9724181958008558, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.6": { "accuracy": 0.9673449019319378, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.7": { "accuracy": 0.9880880096025066, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.8": { "accuracy": 0.9894325943605509, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.9": { "accuracy": 0.9884741210989887, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.10": { "accuracy": 0.9875325196044287, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.11": { "accuracy": 0.9882755563594401, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.12": { "accuracy": 0.9881337507104035, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.13": { "accuracy": 0.9875854060665006, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.14": { "accuracy": 0.9870891972677782, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.15": { "accuracy": 0.9867504658468533, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.16": { "accuracy": 0.9801667189749423, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.17": { "accuracy": 0.990139056724729, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.18": { "accuracy": 0.9891062806273112, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.19": { "accuracy": 0.9861994145030621, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.20": { "accuracy": 0.9865079082956072, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.21": { "accuracy": 0.9833653933892492, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.22": { "accuracy": 0.9856840915745124, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.23": { "accuracy": 0.9849393175099976, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.24": { "accuracy": 0.9839110090688337, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.25": { "accuracy": 0.9849444307328667, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.26": { "accuracy": 0.9809935875528026, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.27": { "accuracy": 0.9842479661456309, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.28": { "accuracy": 0.9828384715074208, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.29": { "accuracy": 0.983858243591385, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.30": { "accuracy": 0.9823532946174964, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.31": { "accuracy": 0.9828728725260589, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.32": { "accuracy": 0.9822730309679173, "total_bits": 832045056.0, "o_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 32 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.33": { "accuracy": 0.9848659632261842, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.34": { "accuracy": 0.9992914507165551, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } }, "model.layers.35": { "accuracy": 0.9989371998235583, "total_bits": 786825216.0, "o_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "down_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "q_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "k_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "v_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "gate_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 }, "up_proj": { "group_size": { "4": 128 }, "bits": [ 4 ], "bits_prop": [ 1 ], "scale_bits": 4, "scale_groups:": 32 } } } }