diff --git "a/.quant_summary.txt" "b/.quant_summary.txt" new file mode 100644--- /dev/null +++ "b/.quant_summary.txt" @@ -0,0 +1,17824 @@ +model.encoder.language_model.layers.0.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.0.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.92e+02 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.0.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.1.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.86e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.70e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.1.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.2.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.2.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.3.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.78e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.80e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.3.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.4.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.60e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.4.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.5.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.82e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.5.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.6.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.00e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.6.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.7.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.46e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.7.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.8.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.8.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.9.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.19e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.9.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.10.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e+02 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.10.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.11.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+02 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.11.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.12.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.72e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.49e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.44e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.12.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.13.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.91e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+02 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.13.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.14.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.31e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.14.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.15.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.02e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.46e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.15.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.16.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.16.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.17.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.49e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.51e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.56e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.17.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.18.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.26e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.63e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.18.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.19.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.34e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.57e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.19.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.20.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.35e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.46e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.20.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.21.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.43e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.35e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.26e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.35e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.21.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.22.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.22.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.23.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.23.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.24.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.24.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.25.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.25.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.26.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.31e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e+01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.53e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.19e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.26.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.27.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.75e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.39e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.27.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.28.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.80e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.43e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.28.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.router.proj.input_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.router.proj.output_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.router.proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.language_model.layers.29.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.78e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.51e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.67e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.56e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.encoder.language_model.layers.29.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.encoder.vision_tower.patch_embedder.input_proj.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.patch_embedder.input_proj.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.patch_embedder.input_proj.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.0.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.0.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.0.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.1.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.1.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.1.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.2.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.2.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.2.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.3.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.3.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.3.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.4.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.4.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.4.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.5.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.5.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.5.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.6.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.6.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.6.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.7.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.7.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.7.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.8.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.8.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.8.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.9.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.9.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.9.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.10.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.10.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.10.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.11.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.11.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.11.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.12.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.12.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.12.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.13.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.13.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.13.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.14.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.14.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.14.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.15.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.15.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.15.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.16.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.16.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.16.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.17.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.17.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.17.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.18.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.18.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.18.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.19.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.19.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.19.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.20.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.20.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.20.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.21.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.21.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.21.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.22.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.22.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.22.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.23.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.23.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.23.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.24.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.24.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.24.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.25.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.25.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.25.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.q_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.q_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.q_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.k_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.k_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.k_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.v_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.v_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.v_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.o_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.o_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.o_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.26.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.encoder.vision_tower.encoder.layers.26.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.gate_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.gate_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.gate_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.up_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.up_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.up_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.down_proj.linear.input_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.down_proj.linear.output_quantizer TensorQuantizer(disabled) +model.encoder.vision_tower.encoder.layers.26.mlp.down_proj.linear.weight_quantizer TensorQuantizer(disabled) +model.encoder.embed_vision.embedding_projection.input_quantizer TensorQuantizer(disabled) +model.encoder.embed_vision.embedding_projection.output_quantizer TensorQuantizer(disabled) +model.encoder.embed_vision.embedding_projection.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.0.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.0.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.1.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.1.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.2.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.2.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.3.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.3.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.4.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.25e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.4.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.5.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.5.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.5.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.5.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.6.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.6.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.6.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.25e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.6.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.7.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.7.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.7.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.92e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.7.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.8.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.8.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.8.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.8.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.9.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.9.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.9.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.69e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.19e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.9.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.10.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.10.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.10.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.95e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.10.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.11.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.11.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.11.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.51e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.68e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.11.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.12.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.12.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.12.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.05e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.49e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.44e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.12.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.13.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.13.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.13.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.13.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.14.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.14.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.14.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.56e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.68e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.14.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.15.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.15.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.15.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.94e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.68e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.46e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.15.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.16.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.16.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.16.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.16.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.17.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.17.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.17.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.49e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.51e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.31e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.56e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.17.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.18.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.18.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.18.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.26e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.63e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.28e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.18.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.19.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.19.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.19.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.84e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.34e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.57e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.19.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.20.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.20.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.20.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.33e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.30e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.42e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.37e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.35e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.46e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.20.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.21.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.21.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.21.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.29e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.43e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.35e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.32e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.24e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.26e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.25e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.41e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.27e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.35e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.21.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.22.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.22.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.22.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.56e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.36e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.17e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.22.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.23.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.23.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.23.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.23.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.24.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.24.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.24.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.09e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.13e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.07e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.11e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.24.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.25.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.25.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.25.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.25.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.26.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.26.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.26.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.64e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.39e+01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.53e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.19e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.10e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.26.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.27.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.27.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.27.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.84e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.39e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.18e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.97e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.27.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.v_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.v_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.v_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.28.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.28.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.28.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.55e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.08e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.35e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.78e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.43e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.37e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.98e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.57e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.43e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.74e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.94e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.48e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.23e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.06e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.41e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.47e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.17e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.28.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.self_attn.q_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.q_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.q_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.k_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.k_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.k_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.o_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.o_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.o_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.q_bmm_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.self_attn.k_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.self_attn.v_bmm_quantizer TensorQuantizer((4, 3) bit fake per-tensor amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.self_attn.softmax_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.mlp.down_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.router.proj.input_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.router.proj.output_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.router.proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.layers.29.experts.gate_up_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.13e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.51e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.69e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.76e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.67e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.60e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.26e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.75e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.49e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.99e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.03e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.04e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.31e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.11e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.00e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.39e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.06e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.71e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.82e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.07e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.14e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.86e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.79e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.29e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=3.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=2.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.59e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.gate_up_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.63e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_input_quantizer TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=4.66e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.0 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.1 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.2 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.32e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.3 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.4 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.5 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.6 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.70e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.7 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.8 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.25e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.9 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.15e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.10 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.96e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.11 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.12 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.13 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.14 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.68e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.15 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.16 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.17 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.22e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.18 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.19 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.72e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.20 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.20e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.21 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.22 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.23 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.45e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.24 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.25 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.26 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.27 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.28 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.29 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.30 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.56e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.31 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.32 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.33 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.34 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.53e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.35 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.36 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.44e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.37 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.38 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.39 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.90e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.40 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.41 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.42 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.43 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.33e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.44 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.64e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.45 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.09e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.46 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.50e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.47 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.48 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.03e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.49 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.50 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.52e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.51 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.52 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.53 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.54 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.27e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.55 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.56 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.57 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.46e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.58 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.16e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.59 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.01e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.60 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.61 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.04e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.62 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=dynamic calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.63 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.65e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.64 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.14e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.65 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.66 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.67 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.68 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.58e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.69 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.48e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.70 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.71 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.18e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.72 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.73 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.74 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.75 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.76 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.30e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.77 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.20e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.78 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.23e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.79 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.80 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.62e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.81 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.66e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.82 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.10e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.83 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.61e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.84 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.01e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.85 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.73e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.86 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.92e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.87 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.05e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.88 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.15e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.89 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.87e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.90 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.00e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.91 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.05e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.92 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.77e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.93 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.28e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.94 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.42e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.95 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.95e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.96 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.22e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.97 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.21e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.98 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.16e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.99 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.08e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.100 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.101 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.91e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.102 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.40e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.103 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.81e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.104 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.105 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.93e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.106 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=9.02e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.107 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.02e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.108 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.85e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.109 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.83e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.110 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.111 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.45e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.112 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.12e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.113 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.114 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.89e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.115 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.38e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.116 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.54e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.117 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.19e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.118 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.84e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.119 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.120 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.121 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=5.55e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.122 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.36e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.123 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.88e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.124 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=1.21e+00 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.125 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=8.24e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.126 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=7.34e-01 calibrator=MaxCalibrator quant) +model.decoder.layers.29.experts.down_proj_weight_quantizers.127 TensorQuantizer((2, 1) bit fake block_sizes={-1: 16, 'type': 'dynamic', 'scale_bits': (4, 3)}, amax=6.80e-01 calibrator=MaxCalibrator quant) +model.decoder.self_conditioning.gate_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.gate_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.gate_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.up_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.up_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.up_proj.weight_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.down_proj.input_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.down_proj.output_quantizer TensorQuantizer(disabled) +model.decoder.self_conditioning.down_proj.weight_quantizer TensorQuantizer(disabled) +lm_head.input_quantizer TensorQuantizer(disabled) +lm_head.output_quantizer TensorQuantizer(disabled) +lm_head.weight_quantizer TensorQuantizer(disabled) +17823 TensorQuantizers found in model