{ "measurement": { "model.layers.0": { "accuracy": 0.9287314414978027, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.1": { "accuracy": 0.914562463760376, "total_bits": 472429600, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.1, 0.9 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.1, 0.9 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.1, 0.9 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.1, 0.9 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.1, 0.9 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.2": { "accuracy": 0.9650976061820984, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.3": { "accuracy": 0.9562180042266846, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.4": { "accuracy": 0.9498416185379028, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.5": { "accuracy": 0.9452450275421143, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.6": { "accuracy": 0.93950355052948, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.7": { "accuracy": 0.9338642358779907, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.8": { "accuracy": 0.9276525974273682, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.9": { "accuracy": 0.9260540008544922, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.10": { "accuracy": 0.9225144386291504, "total_bits": 424064032, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.01, 0.99 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 } }, "model.layers.11": { "accuracy": 0.9225969314575195, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.12": { "accuracy": 0.9214274883270264, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.13": { "accuracy": 0.9215571880340576, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.14": { "accuracy": 0.9176080226898193, "total_bits": 435991584, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 } }, "model.layers.15": { "accuracy": 0.923285961151123, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.16": { "accuracy": 0.9177236557006836, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.17": { "accuracy": 0.9207503795623779, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.18": { "accuracy": 0.9209573268890381, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.19": { "accuracy": 0.9213593006134033, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.20": { "accuracy": 0.920569658279419, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.21": { "accuracy": 0.9229907989501953, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.22": { "accuracy": 0.9201843738555908, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.23": { "accuracy": 0.9208340644836426, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.24": { "accuracy": 0.9214305877685547, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.25": { "accuracy": 0.919379472732544, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.26": { "accuracy": 0.9194035530090332, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.27": { "accuracy": 0.9180495738983154, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.28": { "accuracy": 0.9156413078308105, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } }, "model.layers.29": { "accuracy": 0.9198658466339111, "total_bits": 478983200, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.4, 0.6 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.05, 0.95 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.4, 0.6 ], "scale_bits": 4 } }, "model.layers.30": { "accuracy": 0.9203238487243652, "total_bits": 578335776, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.35, 0.65 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.35, 0.65 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.5, 0.5 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.35, 0.65 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.35, 0.65 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.35, 0.65 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.5, 0.5 ], "scale_bits": 4 } }, "model.layers.31": { "accuracy": 0.9024977684020996, "total_bits": 446084128, "q_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "k_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "v_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 }, "o_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "up_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "gate_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.02, 0.98 ], "scale_bits": 4 }, "down_proj": { "group_size": { "4": 128, "2": 128 }, "bits": [ 4, 2 ], "bits_prop": [ 0.2, 0.8 ], "scale_bits": 4 } } } }