| { |
| "measurement": { |
| "model.layers.0": { |
| "accuracy": 0.9627418518066406, |
| "total_bits": 1122508800.0, |
| "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.1": { |
| "accuracy": 0.988224983215332, |
| "total_bits": 1122508800.0, |
| "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.2": { |
| "accuracy": 0.9828853607177734, |
| "total_bits": 1122508800.0, |
| "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.3": { |
| "accuracy": 0.9805974960327148, |
| "total_bits": 1122508800.0, |
| "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.4": { |
| "accuracy": 0.9719228744506836, |
| "total_bits": 1122508800.0, |
| "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.5": { |
| "accuracy": 0.9817636013031006, |
| "total_bits": 1122508800.0, |
| "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.6": { |
| "accuracy": 0.9828321933746338, |
| "total_bits": 1122508800.0, |
| "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.7": { |
| "accuracy": 0.9890775680541992, |
| "total_bits": 1122508800.0, |
| "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": 0.9881250858306885, |
| "total_bits": 1122508800.0, |
| "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": 0.9886119365692139, |
| "total_bits": 1122508800.0, |
| "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": 0.9889421463012695, |
| "total_bits": 1122508800.0, |
| "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": 0.987459659576416, |
| "total_bits": 1122508800.0, |
| "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": 0.9875621795654297, |
| "total_bits": 1122508800.0, |
| "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": 0.9868230819702148, |
| "total_bits": 1122508800.0, |
| "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": 0.9865565299987793, |
| "total_bits": 1122508800.0, |
| "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": 0.9852416515350342, |
| "total_bits": 1122508800.0, |
| "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": 0.985353946685791, |
| "total_bits": 1122508800.0, |
| "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.17": { |
| "accuracy": 0.984795331954956, |
| "total_bits": 1122508800.0, |
| "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": 0.9835095405578613, |
| "total_bits": 1122508800.0, |
| "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": 0.9838428497314453, |
| "total_bits": 1122508800.0, |
| "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": 0.982266902923584, |
| "total_bits": 1122508800.0, |
| "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": 0.9830894470214844, |
| "total_bits": 1122508800.0, |
| "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": 0.9808683395385742, |
| "total_bits": 1122508800.0, |
| "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.23": { |
| "accuracy": 0.984468936920166, |
| "total_bits": 1122508800.0, |
| "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.24": { |
| "accuracy": 0.9841346740722656, |
| "total_bits": 1122508800.0, |
| "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.25": { |
| "accuracy": 0.9833474159240723, |
| "total_bits": 1122508800.0, |
| "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.26": { |
| "accuracy": 0.9836993217468262, |
| "total_bits": 1122508800.0, |
| "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": 0.9824638366699219, |
| "total_bits": 1122508800.0, |
| "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.28": { |
| "accuracy": 0.9820375442504883, |
| "total_bits": 1122508800.0, |
| "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.29": { |
| "accuracy": 0.9805641174316406, |
| "total_bits": 1122508800.0, |
| "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.30": { |
| "accuracy": 0.9803977012634277, |
| "total_bits": 1122508800.0, |
| "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.31": { |
| "accuracy": 0.9816937446594238, |
| "total_bits": 1122508800.0, |
| "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.32": { |
| "accuracy": 0.98065185546875, |
| "total_bits": 1122508800.0, |
| "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.33": { |
| "accuracy": 0.9798593521118164, |
| "total_bits": 1122508800.0, |
| "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": 0.9807953834533691, |
| "total_bits": 1122508800.0, |
| "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": 0.9799060821533203, |
| "total_bits": 1122508800.0, |
| "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.36": { |
| "accuracy": 0.9804143905639648, |
| "total_bits": 1122508800.0, |
| "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.37": { |
| "accuracy": 0.9817371368408203, |
| "total_bits": 1122508800.0, |
| "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.38": { |
| "accuracy": 0.9827446937561035, |
| "total_bits": 1122508800.0, |
| "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.39": { |
| "accuracy": 0.9829044342041016, |
| "total_bits": 1122508800.0, |
| "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.40": { |
| "accuracy": 0.9842257499694824, |
| "total_bits": 1122508800.0, |
| "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.41": { |
| "accuracy": 0.9848818778991699, |
| "total_bits": 1122508800.0, |
| "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.42": { |
| "accuracy": 0.9853067398071289, |
| "total_bits": 1122508800.0, |
| "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.43": { |
| "accuracy": 0.9829626083374023, |
| "total_bits": 1122508800.0, |
| "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.44": { |
| "accuracy": 0.9822745323181152, |
| "total_bits": 1122508800.0, |
| "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.45": { |
| "accuracy": 0.9821968078613281, |
| "total_bits": 1122508800.0, |
| "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.46": { |
| "accuracy": 0.999725341796875, |
| "total_bits": 1122508800.0, |
| "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.47": { |
| "accuracy": 0.9996490478515625, |
| "total_bits": 1122508800.0, |
| "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 |
| } |
| } |
| } |
| } |