checkpoints/shallow010_naslarge_2b8b_wSAM/statistics_training_checkpoint

Top:
   epoch
   extras
   state_dict
   arch
-------------------------------------
arch: ai85nascifarnet
-------------------------------------
extras: None
-------------------------------------
state_dict:
   conv1_1
     output_shift:         [-1.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [8.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 128 , [128]
        # of unique elements:       128
        min, max, mean: -0.11356868 ,  0.091928445 ,  0.005726372
     weight
        total # of elements, shape: 3456 , [128, 3, 3, 3]
        # of unique elements:       3456
        min, max, mean: -0.7470972 ,  0.8304557 ,  -0.0008773367
   conv1_2
     output_shift:         [-1.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 128 , [128]
        # of unique elements:       128
        min, max, mean: -0.41145262 ,  0.30299085 ,  0.11826816
     weight
        total # of elements, shape: 16384 , [128, 128, 1, 1]
        # of unique elements:       16366
        min, max, mean: -0.5303318 ,  0.5205089 ,  -0.009956859
   conv1_3
     output_shift:         [-3.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 256 , [256]
        # of unique elements:       256
        min, max, mean: -0.72582525 ,  1.0370457 ,  0.08059071
     weight
        total # of elements, shape: 294912 , [256, 128, 3, 3]
        # of unique elements:       294293
        min, max, mean: -0.20158304 ,  0.16796228 ,  -0.0024090037
   conv2_1
     output_shift:         [-4.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 128 , [128]
        # of unique elements:       128
        min, max, mean: -0.49013856 ,  0.5312471 ,  0.1016065
     weight
        total # of elements, shape: 294912 , [128, 256, 3, 3]
        # of unique elements:       294202
        min, max, mean: -0.08584305 ,  0.060128834 ,  -0.00024059539
   conv2_2
     output_shift:         [-1.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 128 , [128]
        # of unique elements:       128
        min, max, mean: -0.6671736 ,  0.7120095 ,  0.08861585
     weight
        total # of elements, shape: 16384 , [128, 128, 1, 1]
        # of unique elements:       16371
        min, max, mean: -0.56641173 ,  0.64063776 ,  -0.016870284
   conv3_1
     output_shift:         [-4.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 64 , [64]
        # of unique elements:       64
        min, max, mean: -0.8250142 ,  0.74603933 ,  -0.0037405742
     weight
        total # of elements, shape: 73728 , [64, 128, 3, 3]
        # of unique elements:       73571
        min, max, mean: -0.10138782 ,  0.11076212 ,  0.00093133084
   conv3_2
     output_shift:         [-1.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 256 , [256]
        # of unique elements:       256
        min, max, mean: -0.33974662 ,  0.44110784 ,  0.07637683
     weight
        total # of elements, shape: 16384 , [256, 64, 1, 1]
        # of unique elements:       16366
        min, max, mean: -0.85561633 ,  0.80795157 ,  -0.020401143
   conv4_1
     output_shift:         [-4.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 128 , [128]
        # of unique elements:       128
        min, max, mean: -0.499701 ,  0.9950852 ,  0.029063346
     weight
        total # of elements, shape: 294912 , [128, 256, 3, 3]
        # of unique elements:       294007
        min, max, mean: -0.09003077 ,  0.09095405 ,  -0.00023980369
   conv4_2
     output_shift:         [-3.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 64 , [64]
        # of unique elements:       64
        min, max, mean: -0.23323293 ,  0.33829257 ,  0.089144416
     weight
        total # of elements, shape: 73728 , [64, 128, 3, 3]
        # of unique elements:       73565
        min, max, mean: -0.23597741 ,  0.21488984 ,  -0.0011587992
   conv5_1
     output_shift:         [-1.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [2.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 128 , [128]
        # of unique elements:       128
        min, max, mean: -0.28117087 ,  0.44147873 ,  0.086634256
     weight
        total # of elements, shape: 8192 , [128, 64, 1, 1]
        # of unique elements:       8119
        min, max, mean: -0.5141475 ,  0.8037016 ,  0.0038255197
   fc
     output_shift:         [1.]
     adjust_output_shift:  [1.]
     quantize_activation:  [1.]
     shift_quantile:       [0.985]
     weight bits:          [8.]
     bias_bits:            [8.]
     bias
        total # of elements, shape: 100 , [100]
        # of unique elements:       100
        min, max, mean: -0.2644152 ,  0.3170493 ,  -0.002889828
     weight
        total # of elements, shape: 51200 , [100, 512]
        # of unique elements:       51181
        min, max, mean: -3.9576318 ,  1.9540635 ,  -0.22963743