From 86afd5f91edbfaa8ba0dab643816bff22d8b2f6a Mon Sep 17 00:00:00 2001 From: nickfraser Date: Thu, 12 Sep 2024 15:53:25 +0100 Subject: [PATCH] notebooks: rerun notebooks. (#1026) --- ...1_quant_tensor_quant_conv2d_overview.ipynb | 95 ++++++++++--------- notebooks/02_quant_activation_overview.ipynb | 53 ++++++----- notebooks/03_anatomy_of_a_quantizer.ipynb | 52 +++++----- notebooks/Brevitas_TVMCon2021.ipynb | 57 ++++++----- notebooks/ONNX_export_tutorial.ipynb | 8 +- notebooks/minifloat_mx_tutorial.ipynb | 26 ++--- notebooks/quantized_recurrent.ipynb | 46 ++++----- 7 files changed, 176 insertions(+), 161 deletions(-) diff --git a/notebooks/01_quant_tensor_quant_conv2d_overview.ipynb b/notebooks/01_quant_tensor_quant_conv2d_overview.ipynb index c0ee56321..e418439e9 100644 --- a/notebooks/01_quant_tensor_quant_conv2d_overview.ipynb +++ b/notebooks/01_quant_tensor_quant_conv2d_overview.ipynb @@ -45,7 +45,8 @@ " dtype: Optional[torch.dtype] = None,\n", " **kwargs) -> None:\n", " # avoid an init error in the super class by setting padding to 0\n", - " if padding_mode == 'zeros' and padding == 'same' and stride > 1:\n", + " if padding_mode == 'zeros' and padding == 'same' and (stride > 1 if isinstance(\n", + " stride, int) else any(map(lambda x: x > 1, stride))):\n", " padding = 0\n", " is_same_padded_strided = True\n", " else:\n", @@ -166,9 +167,10 @@ "name": "stderr", "output_type": "stream", "text": [ - "/home/giuseppe/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/_tensor.py:1394: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1708025842427/work/c10/core/TensorImpl.h:1908.)\n", - " return super().rename(names)\n", - "/home/giuseppe/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/conv.py:456: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1708025842427/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/_tensor.py:1255: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/c10/core/TensorImpl.h:1758.)\n", + " return super(Tensor, self).rename(names)\n", + "[W NNPACK.cpp:53] Could not initialize NNPACK! Reason: Unsupported hardware.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/conv.py:459: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " return F.conv2d(input, weight, bias, self.stride,\n" ] }, @@ -514,7 +516,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/tmp/ipykernel_528161/1377665000.py:1: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1708025842427/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/tmp/ipykernel_81376/1377665000.py:1: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " torch.tanh(quant_tensor)\n" ] }, @@ -632,17 +634,17 @@ { "data": { "text/plain": [ - "IntQuantTensor(value=tensor([[[[-9.9000e-03, -7.1000e-03, -4.7000e-03, 5.0000e-03, -1.2300e-02],\n", - " [-8.2000e-03, 8.5000e-03, -1.2000e-03, -1.2500e-02, 4.4000e-03],\n", - " [ 4.3000e-03, -6.3000e-03, -9.4000e-03, 1.0400e-02, -1.2100e-02],\n", - " [ 1.1700e-02, -3.6000e-03, 5.3000e-03, -1.1700e-02, -4.3000e-03],\n", - " [-8.8000e-03, 1.0900e-02, -8.3000e-03, -2.9000e-03, 1.2400e-02]],\n", + "IntQuantTensor(value=tensor([[[[ 7.2000e-03, -3.7000e-03, 7.7000e-03, -2.4000e-03, -8.9000e-03],\n", + " [-1.2000e-02, -8.1000e-03, 7.2000e-03, -1.1300e-02, -9.7000e-03],\n", + " [-1.0000e-03, 1.0100e-02, 3.8000e-03, -1.1900e-02, 6.9000e-03],\n", + " [ 8.3000e-03, 1.0000e-04, -6.9000e-03, 3.9000e-03, -5.4000e-03],\n", + " [ 1.1300e-02, -6.0000e-03, 9.7000e-03, 0.0000e+00, 1.0900e-02]],\n", "\n", - " [[ 9.3000e-03, -8.5000e-03, 6.5000e-03, -2.7000e-03, -3.4000e-03],\n", - " [-1.0000e-04, -1.1000e-02, 8.3000e-03, 1.9000e-03, -9.8000e-03],\n", - " [ 4.3000e-03, -8.5000e-03, 1.1000e-02, 5.3000e-03, 3.4000e-03],\n", - " [ 8.1000e-03, 9.8000e-03, 6.8000e-03, 1.5000e-03, 6.3000e-03],\n", - " [ 5.7000e-03, -8.5000e-03, 5.2000e-03, -3.0000e-04, 4.9000e-03]]]]), scale=tensor(1.0000e-04), zero_point=tensor(0.), bit_width=tensor(8.), signed_t=tensor(True), training_t=tensor(True))" + " [[-1.0900e-02, 1.1400e-02, -6.4000e-03, 9.2000e-03, 7.1000e-03],\n", + " [-6.0000e-04, 9.2000e-03, -8.5000e-03, 5.0000e-03, 6.5000e-03],\n", + " [-8.3000e-03, -1.2000e-03, 7.4000e-03, 9.2000e-03, -6.0000e-04],\n", + " [-2.1000e-03, 9.5000e-03, 3.0000e-04, -2.9000e-03, -6.5000e-03],\n", + " [-1.1800e-02, -4.8000e-03, 5.4000e-03, -2.5000e-03, 9.0000e-04]]]]), scale=tensor(1.0000e-04), zero_point=tensor(0.), bit_width=tensor(8.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 17, @@ -702,7 +704,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.0019, 0.0049, -0.0012],\n", + "IntQuantTensor(value=tensor([[[[-0.0019, 0.0049, -0.0012],\n", " [-0.0012, 0.0050, -0.0074],\n", " [-0.0023, -0.0035, -0.0033]],\n", "\n", @@ -712,7 +714,7 @@ "\n", " [[-0.0045, -0.0010, 0.0002],\n", " [-0.0044, 0.0027, 0.0025],\n", - " [-0.0009, 0.0040, -0.0044]]]], grad_fn=), scale=tensor([[[[1.8307e-07]]]], grad_fn=), zero_point=tensor(0.), bit_width=tensor(21.), signed_t=tensor(True), training_t=tensor(True))" + " [-0.0009, 0.0040, -0.0044]]]], grad_fn=), scale=tensor([[[[1.8307e-07]]]], grad_fn=), zero_point=tensor([0.]), bit_width=tensor(21.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 18, @@ -759,7 +761,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.0073, 0.0040, -0.0011],\n", + "IntQuantTensor(value=tensor([[[[-0.0073, 0.0040, -0.0011],\n", " [-0.0033, 0.0078, -0.0028],\n", " [ 0.0005, -0.0025, -0.0008]],\n", "\n", @@ -769,7 +771,7 @@ "\n", " [[-0.0010, 0.0047, 0.0025],\n", " [-0.0014, 0.0021, -0.0039],\n", - " [ 0.0036, -0.0003, 0.0026]]]], grad_fn=), scale=tensor([[[[1.7393e-07]]]], grad_fn=), zero_point=tensor(0.), bit_width=tensor(21.), signed_t=tensor(True), training_t=tensor(True))" + " [ 0.0036, -0.0003, 0.0026]]]], grad_fn=), scale=tensor([[[[1.7393e-07]]]], grad_fn=), zero_point=tensor([0.]), bit_width=tensor(21.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 20, @@ -798,7 +800,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.2117, -0.4811, 0.0385],\n", + "IntQuantTensor(value=tensor([[[[-0.2117, -0.4811, 0.0385],\n", " [-0.5100, -0.2502, -0.2213],\n", " [-0.5773, 0.0192, -0.5485]],\n", "\n", @@ -878,11 +880,10 @@ "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", - "Cell \u001b[0;32mIn[41], line 6\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mquant\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mscaled_int\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m Int8Bias\n\u001b[1;32m 3\u001b[0m bias_quant_conv \u001b[38;5;241m=\u001b[39m QuantConv2d(\n\u001b[1;32m 4\u001b[0m in_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m2\u001b[39m, out_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m3\u001b[39m, kernel_size\u001b[38;5;241m=\u001b[39m(\u001b[38;5;241m3\u001b[39m,\u001b[38;5;241m3\u001b[39m), bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m 5\u001b[0m bias_quant\u001b[38;5;241m=\u001b[39mInt8Bias, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 6\u001b[0m \u001b[43mbias_quant_conv\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrandn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m1\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m2\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1509\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1511\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1518\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1519\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1520\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 1523\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n", - "File \u001b[0;32m~/Documents/git/brevitas/src/brevitas/nn/quant_conv.py:198\u001b[0m, in \u001b[0;36mQuantConv2d.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 197\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[0;32m--> 198\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/Documents/git/brevitas/src/brevitas/nn/quant_layer.py:173\u001b[0m, in \u001b[0;36mQuantWeightBiasInputOutputLayer.forward_impl\u001b[0;34m(self, inp)\u001b[0m\n\u001b[1;32m 169\u001b[0m compute_output_quant_tensor \u001b[38;5;241m=\u001b[39m \u001b[38;5;28misinstance\u001b[39m(quant_input, QuantTensor) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(\n\u001b[1;32m 170\u001b[0m quant_weight, QuantTensor)\n\u001b[1;32m 171\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (compute_output_quant_tensor \u001b[38;5;129;01mor\u001b[39;00m\n\u001b[1;32m 172\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mis_output_quant_enabled) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreturn_quant_tensor:\n\u001b[0;32m--> 173\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mQuantLayer is not correctly configured\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 175\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant_input, QuantTensor) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant_weight, QuantTensor):\n\u001b[1;32m 176\u001b[0m output_bit_width \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mmax_acc_bit_width(quant_input\u001b[38;5;241m.\u001b[39mbit_width, quant_weight\u001b[38;5;241m.\u001b[39mbit_width)\n", + "Cell \u001b[0;32mIn[23], line 6\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mquant\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mscaled_int\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m Int8Bias\n\u001b[1;32m 3\u001b[0m bias_quant_conv \u001b[38;5;241m=\u001b[39m QuantConv2d(\n\u001b[1;32m 4\u001b[0m in_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m2\u001b[39m, out_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m3\u001b[39m, kernel_size\u001b[38;5;241m=\u001b[39m(\u001b[38;5;241m3\u001b[39m,\u001b[38;5;241m3\u001b[39m), bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m 5\u001b[0m bias_quant\u001b[38;5;241m=\u001b[39mInt8Bias, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 6\u001b[0m \u001b[43mbias_quant_conv\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrandn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m1\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m2\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1194\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m 1190\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1191\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1192\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1193\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1195\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m 1196\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_conv.py:194\u001b[0m, in \u001b[0;36mQuantConv2d.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 193\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[0;32m--> 194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_layer.py:152\u001b[0m, in \u001b[0;36mQuantWeightBiasInputOutputLayer.forward_impl\u001b[0;34m(self, inp)\u001b[0m\n\u001b[1;32m 148\u001b[0m compute_output_quant_tensor \u001b[38;5;241m=\u001b[39m \u001b[38;5;28misinstance\u001b[39m(quant_input, QuantTensor) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(\n\u001b[1;32m 149\u001b[0m quant_weight, QuantTensor)\n\u001b[1;32m 150\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (compute_output_quant_tensor \u001b[38;5;129;01mor\u001b[39;00m\n\u001b[1;32m 151\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39moutput_quant\u001b[38;5;241m.\u001b[39mis_quant_enabled) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreturn_quant_tensor:\n\u001b[0;32m--> 152\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mQuantLayer is not correctly configured\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 154\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[1;32m 155\u001b[0m quant_bias \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias_quant(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias, quant_input, quant_weight)\n", "\u001b[0;31mRuntimeError\u001b[0m: QuantLayer is not correctly configured" ] } @@ -911,18 +912,18 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-2.4238e-03, -5.6598e-03, 5.1882e-03],\n", + "IntQuantTensor(value=tensor([[[[-2.4238e-03, -5.6598e-03, 5.1882e-03],\n", " [-6.5582e-03, 8.9274e-03, 4.9640e-04],\n", " [ 9.6283e-03, -1.7466e-03, -4.8311e-03]],\n", "\n", " [[ 2.9322e-03, -3.1358e-03, -6.2727e-04],\n", - " [ 2.8722e-06, -3.7981e-03, 1.0973e-02],\n", + " [ 2.8723e-06, -3.7981e-03, 1.0973e-02],\n", " [-4.1031e-03, 6.5909e-03, -4.2369e-03]],\n", "\n", " [[ 4.1967e-03, -7.0733e-03, 1.6456e-03],\n", " [ 1.8197e-03, -3.1683e-03, 4.8200e-03],\n", " [-3.2585e-04, 3.1055e-03, 1.9703e-03]]]],\n", - " grad_fn=), scale=tensor([[[[1.7953e-07]]]], grad_fn=), zero_point=tensor(0.), bit_width=tensor(22.), signed_t=tensor(True), training_t=tensor(True))" + " grad_fn=), scale=tensor([[[[1.7953e-07]]]], grad_fn=), zero_point=tensor([0.]), bit_width=tensor(22.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 24, @@ -949,7 +950,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.2816, -0.5271, -0.1748],\n", + "IntQuantTensor(value=tensor([[[[-0.2816, -0.5271, -0.1748],\n", " [-0.4247, -0.1575, 0.0681],\n", " [ 0.6528, -0.5346, -0.0657]],\n", "\n", @@ -959,7 +960,7 @@ "\n", " [[ 0.5586, 0.0665, -0.5807],\n", " [ 0.5565, 0.1780, -0.0555],\n", - " [-0.1080, 0.0791, -0.2262]]]], grad_fn=), scale=tensor([[[[4.2009e-05]]]], grad_fn=), zero_point=tensor(0.), bit_width=tensor(22.), signed_t=tensor(True), training_t=tensor(True))" + " [-0.1080, 0.0791, -0.2262]]]], grad_fn=), scale=tensor([[[[4.2009e-05]]]], grad_fn=), zero_point=tensor([0.]), bit_width=tensor(22.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 25, @@ -982,7 +983,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.0058, 0.0030, 0.0030],\n", + "IntQuantTensor(value=tensor([[[[-0.0058, 0.0030, 0.0030],\n", " [-0.0013, -0.0002, 0.0043],\n", " [-0.0061, 0.0033, -0.0001]],\n", "\n", @@ -992,7 +993,7 @@ "\n", " [[-0.0061, 0.0053, -0.0004],\n", " [ 0.0028, 0.0031, -0.0037],\n", - " [ 0.0027, -0.0048, -0.0044]]]], grad_fn=), scale=tensor([[[[1.7370e-07]]]], grad_fn=), zero_point=tensor(0.), bit_width=tensor(22.), signed_t=tensor(True), training_t=tensor(True))" + " [ 0.0027, -0.0048, -0.0044]]]], grad_fn=), scale=tensor([[[[1.7370e-07]]]], grad_fn=), zero_point=tensor([0.]), bit_width=tensor(22.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 26, @@ -1022,6 +1023,14 @@ ] }, "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/parameter_quant.py:154: UserWarning: No quant bias cache found, set cache_inference_quant_bias=True and run an inference pass first\n", + " warn(\n" + ] + }, { "ename": "RuntimeError", "evalue": "Input scale required", @@ -1029,14 +1038,12 @@ "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", - "Cell \u001b[0;32mIn[45], line 4\u001b[0m\n\u001b[1;32m 1\u001b[0m output_bias_quant_conv \u001b[38;5;241m=\u001b[39m QuantConv2d(\n\u001b[1;32m 2\u001b[0m in_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m2\u001b[39m, out_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m3\u001b[39m, kernel_size\u001b[38;5;241m=\u001b[39m(\u001b[38;5;241m3\u001b[39m,\u001b[38;5;241m3\u001b[39m), bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m 3\u001b[0m output_quant\u001b[38;5;241m=\u001b[39mInt8ActPerTensorFloat, bias_quant\u001b[38;5;241m=\u001b[39mInt8Bias, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 4\u001b[0m \u001b[43moutput_bias_quant_conv\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrandn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m1\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m2\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1509\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1511\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1518\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1519\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1520\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 1523\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n", - "File \u001b[0;32m~/Documents/git/brevitas/src/brevitas/nn/quant_conv.py:198\u001b[0m, in \u001b[0;36mQuantConv2d.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 197\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[0;32m--> 198\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/Documents/git/brevitas/src/brevitas/nn/quant_layer.py:181\u001b[0m, in \u001b[0;36mQuantWeightBiasInputOutputLayer.forward_impl\u001b[0;34m(self, inp)\u001b[0m\n\u001b[1;32m 178\u001b[0m output_signed \u001b[38;5;241m=\u001b[39m quant_input\u001b[38;5;241m.\u001b[39msigned \u001b[38;5;129;01mor\u001b[39;00m quant_weight\u001b[38;5;241m.\u001b[39msigned\n\u001b[1;32m 180\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[0;32m--> 181\u001b[0m quant_bias \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias_quant\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43moutput_scale\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 182\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtraining \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcache_inference_quant_bias \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant_bias,\n\u001b[1;32m 183\u001b[0m QuantTensor):\n\u001b[1;32m 184\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias_quant\u001b[38;5;241m.\u001b[39m_cached_bias \u001b[38;5;241m=\u001b[39m _CachedIO(quant_bias\u001b[38;5;241m.\u001b[39mdetach(), metadata_only\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m)\n", - "File \u001b[0;32m~/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1509\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1511\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/torch_2.1/lib/python3.11/site-packages/torch/nn/modules/module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1518\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1519\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1520\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 1523\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n", - "File \u001b[0;32m~/Documents/git/brevitas/src/brevitas/proxy/parameter_quant.py:240\u001b[0m, in \u001b[0;36mBiasQuantProxyFromInjector.forward\u001b[0;34m(self, x, input_scale)\u001b[0m\n\u001b[1;32m 238\u001b[0m impl \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexport_handler \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexport_mode \u001b[38;5;28;01melse\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtensor_quant\n\u001b[1;32m 239\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrequires_input_scale \u001b[38;5;129;01mand\u001b[39;00m input_scale \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[0;32m--> 240\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mInput scale required\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 242\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrequires_input_scale:\n\u001b[1;32m 243\u001b[0m input_scale \u001b[38;5;241m=\u001b[39m input_scale\u001b[38;5;241m.\u001b[39mview(\u001b[38;5;241m-\u001b[39m\u001b[38;5;241m1\u001b[39m)\n", + "Cell \u001b[0;32mIn[27], line 4\u001b[0m\n\u001b[1;32m 1\u001b[0m output_bias_quant_conv \u001b[38;5;241m=\u001b[39m QuantConv2d(\n\u001b[1;32m 2\u001b[0m in_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m2\u001b[39m, out_channels\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m3\u001b[39m, kernel_size\u001b[38;5;241m=\u001b[39m(\u001b[38;5;241m3\u001b[39m,\u001b[38;5;241m3\u001b[39m), bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m 3\u001b[0m output_quant\u001b[38;5;241m=\u001b[39mInt8ActPerTensorFloat, bias_quant\u001b[38;5;241m=\u001b[39mInt8Bias, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 4\u001b[0m \u001b[43moutput_bias_quant_conv\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrandn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m1\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m2\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m5\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1194\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m 1190\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1191\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1192\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1193\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1195\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m 1196\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_conv.py:194\u001b[0m, in \u001b[0;36mQuantConv2d.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 193\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[0;32m--> 194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_layer.py:155\u001b[0m, in \u001b[0;36mQuantWeightBiasInputOutputLayer.forward_impl\u001b[0;34m(self, inp)\u001b[0m\n\u001b[1;32m 152\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mQuantLayer is not correctly configured\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 154\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[0;32m--> 155\u001b[0m quant_bias \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias_quant\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_input\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_weight\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 156\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 157\u001b[0m quant_bias \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1194\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m 1190\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1191\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1192\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1193\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1195\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m 1196\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/parameter_quant.py:330\u001b[0m, in \u001b[0;36mBiasQuantProxyFromInjector.forward\u001b[0;34m(self, x, input, weight)\u001b[0m\n\u001b[1;32m 328\u001b[0m input_scale \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mscale()\n\u001b[1;32m 329\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m input_scale \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[0;32m--> 330\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mInput scale required\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 331\u001b[0m \u001b[38;5;28;01melif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrequires_input_scale \u001b[38;5;129;01mand\u001b[39;00m input_scale \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mis_quant_enabled:\n\u001b[1;32m 332\u001b[0m input_scale \u001b[38;5;241m=\u001b[39m input_scale\u001b[38;5;241m.\u001b[39mview(\u001b[38;5;241m-\u001b[39m\u001b[38;5;241m1\u001b[39m)\n", "\u001b[0;31mRuntimeError\u001b[0m: Input scale required" ] } @@ -1107,7 +1114,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.6912, 0.0086, 0.1628],\n", + "IntQuantTensor(value=tensor([[[[-0.6912, 0.0086, 0.1628],\n", " [-0.4786, -0.8073, 0.5224],\n", " [ 0.4157, 0.4686, 0.2560]],\n", "\n", @@ -1117,11 +1124,11 @@ "\n", " [[-0.6168, -0.5241, -0.6593],\n", " [ 0.6408, 0.2674, 0.4537],\n", - " [-0.3744, -0.7771, -0.2848]]]], grad_fn=), scale=tensor([[[[3.0094e-05]]]], grad_fn=), zero_point=tensor([[[[ 339.3404]],\n", + " [-0.3744, -0.7771, -0.2848]]]], grad_fn=), scale=tensor([[[[3.0094e-05]]]], grad_fn=), zero_point=tensor([[[[ 339.3406]],\n", "\n", " [[-4597.1797]],\n", "\n", - " [[-3452.3711]]]], grad_fn=), bit_width=tensor(21.), signed_t=tensor(True), training_t=tensor(True))" + " [[-3452.3713]]]], grad_fn=), bit_width=tensor(21.), signed_t=tensor(True), training_t=tensor(True))" ] }, "execution_count": 29, @@ -1218,7 +1225,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.10.13" }, "vscode": { "interpreter": { diff --git a/notebooks/02_quant_activation_overview.ipynb b/notebooks/02_quant_activation_overview.ipynb index 1080b0b0b..962e249e7 100644 --- a/notebooks/02_quant_activation_overview.ipynb +++ b/notebooks/02_quant_activation_overview.ipynb @@ -51,8 +51,11 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/torch/_tensor.py:1362: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449183005/work/c10/core/TensorImpl.h:1900.)\n", - " return super().rename(names)\n" + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/_tensor.py:1255: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/c10/core/TensorImpl.h:1758.)\n", + " return super(Tensor, self).rename(names)\n", + "[W NNPACK.cpp:53] Could not initialize NNPACK! Reason: Unsupported hardware.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/conv.py:459: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", + " return F.conv2d(input, weight, bias, self.stride,\n" ] } ], @@ -163,7 +166,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.4566, -0.5707, -0.5517, 0.5897, 1.5409],\n", + "IntQuantTensor(value=tensor([[[[-0.4566, -0.5707, -0.5517, 0.5897, 1.5409],\n", " [ 0.5136, -0.5897, -0.5707, 0.1902, -0.0761],\n", " [-0.4946, -1.5029, -0.1902, 0.4376, 1.3317],\n", " [-1.6361, 2.0736, 1.7122, 2.3780, -1.1224],\n", @@ -252,7 +255,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.4566, -0.5707, -0.5517, 0.5897, 1.5409],\n", + "IntQuantTensor(value=tensor([[[[-0.4566, -0.5707, -0.5517, 0.5897, 1.5409],\n", " [ 0.5136, -0.5897, -0.5707, 0.1902, -0.0761],\n", " [-0.4946, -1.5029, -0.1902, 0.4376, 1.3317],\n", " [-1.6361, 2.0736, 1.7122, 2.3780, -1.1224],\n", @@ -292,7 +295,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[0.0000, 0.0000, 0.0000, 0.5974, 1.5402],\n", + "IntQuantTensor(value=tensor([[[[0.0000, 0.0000, 0.0000, 0.5974, 1.5402],\n", " [0.5041, 0.0000, 0.0000, 0.1867, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.4481, 1.3255],\n", " [0.0000, 2.0817, 1.7083, 2.3804, 0.0000],\n", @@ -367,17 +370,16 @@ "outputs": [ { "ename": "AssertionError", - "evalue": "", + "evalue": "QuantLayer is not correctly configured, check if warnings were raised", "output_type": "error", "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mAssertionError\u001b[0m Traceback (most recent call last)", - "Cell \u001b[0;32mIn[11], line 4\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mnn\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m QuantSigmoid\n\u001b[1;32m 3\u001b[0m return_disabled_quant_sigmoid \u001b[38;5;241m=\u001b[39m QuantSigmoid(act_quant\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mNone\u001b[39;00m, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 4\u001b[0m sigmoid_out_tensor \u001b[38;5;241m=\u001b[39m return_disabled_quant_sigmoid(out_tensor)\n\u001b[1;32m 5\u001b[0m sigmoid_out_tensor\n", - "File \u001b[0;32m/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/torch/nn/modules/module.py:1518\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1516\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1517\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1518\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/torch/nn/modules/module.py:1527\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1522\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1523\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1524\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1525\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1526\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1527\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m forward_call(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[1;32m 1529\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 1530\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/quant_layer.py:143\u001b[0m, in \u001b[0;36mQuantNonLinearActLayer.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 141\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n\u001b[1;32m 142\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mact_quant(quant_input)\n\u001b[0;32m--> 143\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpack_output(out)\n\u001b[1;32m 144\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:183\u001b[0m, in \u001b[0;36mQuantLayerMixin.pack_output\u001b[0;34m(self, quant_output)\u001b[0m\n\u001b[1;32m 181\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_set_global_is_quant_layer(\u001b[38;5;28;01mFalse\u001b[39;00m)\n\u001b[1;32m 182\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreturn_quant_tensor:\n\u001b[0;32m--> 183\u001b[0m \u001b[38;5;28;01massert\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant_output, QuantTensor)\n\u001b[1;32m 184\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m quant_output\n\u001b[1;32m 185\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n", - "\u001b[0;31mAssertionError\u001b[0m: " + "Cell \u001b[0;32mIn[11], line 4\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mnn\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m QuantSigmoid\n\u001b[1;32m 3\u001b[0m return_disabled_quant_sigmoid \u001b[38;5;241m=\u001b[39m QuantSigmoid(act_quant\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mNone\u001b[39;00m, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 4\u001b[0m sigmoid_out_tensor \u001b[38;5;241m=\u001b[39m \u001b[43mreturn_disabled_quant_sigmoid\u001b[49m\u001b[43m(\u001b[49m\u001b[43mout_tensor\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 5\u001b[0m sigmoid_out_tensor\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1194\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m 1190\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1191\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1192\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1193\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1195\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m 1196\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_layer.py:53\u001b[0m, in \u001b[0;36mQuantNonLinearActLayer.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 51\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n\u001b[1;32m 52\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mact_quant(quant_input)\n\u001b[0;32m---> 53\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mpack_output\u001b[49m\u001b[43m(\u001b[49m\u001b[43mout\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 54\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:97\u001b[0m, in \u001b[0;36mQuantLayerMixin.pack_output\u001b[0;34m(self, quant_output)\u001b[0m\n\u001b[1;32m 95\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mpack_output\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_output: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[1;32m 96\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreturn_quant_tensor:\n\u001b[0;32m---> 97\u001b[0m \u001b[38;5;28;01massert\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant_output, QuantTensor), \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mQuantLayer is not correctly configured, check if warnings were raised\u001b[39m\u001b[38;5;124m'\u001b[39m\n\u001b[1;32m 98\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m quant_output\n\u001b[1;32m 99\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n", + "\u001b[0;31mAssertionError\u001b[0m: QuantLayer is not correctly configured, check if warnings were raised" ] } ], @@ -406,7 +408,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[0.0000, 0.0000, 0.0000, 0.5854, 1.5485],\n", + "IntQuantTensor(value=tensor([[[[0.0000, 0.0000, 0.0000, 0.5854, 1.5485],\n", " [0.5099, 0.0000, 0.0000, 0.1888, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.4532, 1.3219],\n", " [0.0000, 2.0772, 1.6996, 2.3794, 0.0000],\n", @@ -613,16 +615,17 @@ "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mDependencyError\u001b[0m Traceback (most recent call last)", - "Cell \u001b[0;32mIn[18], line 3\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mnn\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m QuantHardTanh\n\u001b[0;32m----> 3\u001b[0m QuantHardTanh()\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/quant_activation.py:96\u001b[0m, in \u001b[0;36mQuantHardTanh.__init__\u001b[0;34m(self, act_quant, input_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 90\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 91\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 92\u001b[0m act_quant: Optional[ActQuantType] \u001b[38;5;241m=\u001b[39m Int8ActPerTensorFloatMinMaxInit,\n\u001b[1;32m 93\u001b[0m input_quant: Optional[ActQuantType] \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m,\n\u001b[1;32m 94\u001b[0m return_quant_tensor: \u001b[38;5;28mbool\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m,\n\u001b[1;32m 95\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m---> 96\u001b[0m QuantNLAL\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 97\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 98\u001b[0m act_impl\u001b[38;5;241m=\u001b[39mnn\u001b[38;5;241m.\u001b[39mHardtanh,\n\u001b[1;32m 99\u001b[0m passthrough_act\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m 100\u001b[0m input_quant\u001b[38;5;241m=\u001b[39minput_quant,\n\u001b[1;32m 101\u001b[0m act_quant\u001b[38;5;241m=\u001b[39mact_quant,\n\u001b[1;32m 102\u001b[0m return_quant_tensor\u001b[38;5;241m=\u001b[39mreturn_quant_tensor,\n\u001b[1;32m 103\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/quant_layer.py:37\u001b[0m, in \u001b[0;36mQuantNonLinearActLayer.__init__\u001b[0;34m(self, act_impl, passthrough_act, input_quant, act_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 35\u001b[0m QuantLayerMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, return_quant_tensor)\n\u001b[1;32m 36\u001b[0m QuantInputMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, input_quant, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[0;32m---> 37\u001b[0m QuantNonLinearActMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, act_impl, passthrough_act, act_quant, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/mixin/act.py:118\u001b[0m, in \u001b[0;36mQuantNonLinearActMixin.__init__\u001b[0;34m(self, act_impl, passthrough_act, act_quant, act_proxy_prefix, act_kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 107\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 108\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 109\u001b[0m act_impl: Optional[Type[Module]],\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 113\u001b[0m act_kwargs_prefix\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m 114\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m 115\u001b[0m prefixed_kwargs \u001b[38;5;241m=\u001b[39m {\n\u001b[1;32m 116\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m: act_impl,\n\u001b[1;32m 117\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mpassthrough_act\u001b[39m\u001b[38;5;124m'\u001b[39m: passthrough_act}\n\u001b[0;32m--> 118\u001b[0m QuantProxyMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 119\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 120\u001b[0m quant\u001b[38;5;241m=\u001b[39mact_quant,\n\u001b[1;32m 121\u001b[0m proxy_prefix\u001b[38;5;241m=\u001b[39mact_proxy_prefix,\n\u001b[1;32m 122\u001b[0m kwargs_prefix\u001b[38;5;241m=\u001b[39mact_kwargs_prefix,\n\u001b[1;32m 123\u001b[0m proxy_protocol\u001b[38;5;241m=\u001b[39mActQuantProxyProtocol,\n\u001b[1;32m 124\u001b[0m none_quant_injector\u001b[38;5;241m=\u001b[39mNoneActQuant,\n\u001b[1;32m 125\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mprefixed_kwargs,\n\u001b[1;32m 126\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:71\u001b[0m, in \u001b[0;36mQuantProxyMixin.__init__\u001b[0;34m(self, quant, proxy_protocol, none_quant_injector, proxy_prefix, kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 69\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant\n\u001b[1;32m 70\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant_injector\u001b[38;5;241m.\u001b[39mlet(\u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mfilter_kwargs(kwargs_prefix, kwargs))\n\u001b[0;32m---> 71\u001b[0m quant \u001b[38;5;241m=\u001b[39m quant_injector\u001b[38;5;241m.\u001b[39mproxy_class(\u001b[38;5;28mself\u001b[39m, quant_injector)\n\u001b[1;32m 72\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 73\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant, proxy_protocol):\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/runtime_quant.py:89\u001b[0m, in \u001b[0;36mActQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 88\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector):\n\u001b[0;32m---> 89\u001b[0m QuantProxyFromInjector\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector)\n\u001b[1;32m 90\u001b[0m ActQuantProxyProtocol\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m)\n\u001b[1;32m 91\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mis_passthrough_act \u001b[38;5;241m=\u001b[39m _is_passthrough_act(quant_injector)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/quant_proxy.py:89\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 87\u001b[0m \u001b[38;5;66;03m# Use a normal list and not a ModuleList since this is a pointer to parent modules\u001b[39;00m\n\u001b[1;32m 88\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list \u001b[38;5;241m=\u001b[39m []\n\u001b[0;32m---> 89\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39madd_tracked_module(quant_layer)\n\u001b[1;32m 90\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdisable_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/quant_proxy.py:131\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.add_tracked_module\u001b[0;34m(self, module)\u001b[0m\n\u001b[1;32m 129\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list\u001b[38;5;241m.\u001b[39mappend(module)\n\u001b[1;32m 130\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mupdate_tracked_modules()\n\u001b[0;32m--> 131\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39minit_tensor_quant()\n\u001b[1;32m 132\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 133\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mTrying to add None as a parent module.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/runtime_quant.py:102\u001b[0m, in \u001b[0;36mActQuantProxyFromInjector.init_tensor_quant\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 101\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21minit_tensor_quant\u001b[39m(\u001b[38;5;28mself\u001b[39m):\n\u001b[0;32m--> 102\u001b[0m tensor_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector\u001b[38;5;241m.\u001b[39mtensor_quant\n\u001b[1;32m 103\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector:\n\u001b[1;32m 104\u001b[0m act_impl \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector\u001b[38;5;241m.\u001b[39mact_impl\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/inject/__init__.py:129\u001b[0m, in \u001b[0;36m_ExtendedInjectorType.__getattr__\u001b[0;34m(cls, attrname)\u001b[0m\n\u001b[1;32m 126\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 127\u001b[0m message \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m can not resolve attribute \u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 128\u001b[0m \u001b[38;5;28mcls\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, current_attr)\n\u001b[0;32m--> 129\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m DependencyError(message)\n\u001b[1;32m 131\u001b[0m marker, attribute, args, have_defaults \u001b[38;5;241m=\u001b[39m spec\n\u001b[1;32m 133\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mset\u001b[39m(args)\u001b[38;5;241m.\u001b[39missubset(cached):\n", + "Cell \u001b[0;32mIn[18], line 3\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mnn\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m QuantHardTanh\n\u001b[0;32m----> 3\u001b[0m \u001b[43mQuantHardTanh\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_activation.py:96\u001b[0m, in \u001b[0;36mQuantHardTanh.__init__\u001b[0;34m(self, act_quant, input_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 90\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 91\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 92\u001b[0m act_quant: Optional[ActQuantType] \u001b[38;5;241m=\u001b[39m Int8ActPerTensorFloatMinMaxInit,\n\u001b[1;32m 93\u001b[0m input_quant: Optional[ActQuantType] \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m,\n\u001b[1;32m 94\u001b[0m return_quant_tensor: \u001b[38;5;28mbool\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m,\n\u001b[1;32m 95\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m---> 96\u001b[0m \u001b[43mQuantNLAL\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m 97\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 98\u001b[0m \u001b[43m \u001b[49m\u001b[43mact_impl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mnn\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mHardtanh\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 99\u001b[0m \u001b[43m \u001b[49m\u001b[43mpassthrough_act\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 100\u001b[0m \u001b[43m \u001b[49m\u001b[43minput_quant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43minput_quant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 101\u001b[0m \u001b[43m \u001b[49m\u001b[43mact_quant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_quant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 102\u001b[0m \u001b[43m \u001b[49m\u001b[43mreturn_quant_tensor\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mreturn_quant_tensor\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 103\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_layer.py:34\u001b[0m, in \u001b[0;36mQuantNonLinearActLayer.__init__\u001b[0;34m(self, act_impl, passthrough_act, input_quant, act_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 32\u001b[0m QuantLayerMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, return_quant_tensor)\n\u001b[1;32m 33\u001b[0m QuantInputMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, input_quant, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[0;32m---> 34\u001b[0m \u001b[43mQuantNonLinearActMixin\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mact_impl\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mpassthrough_act\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mact_quant\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/act.py:66\u001b[0m, in \u001b[0;36mQuantNonLinearActMixin.__init__\u001b[0;34m(self, act_impl, passthrough_act, act_quant, act_proxy_prefix, act_kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 55\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 56\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 57\u001b[0m act_impl: Optional[Type[Module]],\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 61\u001b[0m act_kwargs_prefix\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m 62\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m 63\u001b[0m prefixed_kwargs \u001b[38;5;241m=\u001b[39m {\n\u001b[1;32m 64\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m: act_impl,\n\u001b[1;32m 65\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mpassthrough_act\u001b[39m\u001b[38;5;124m'\u001b[39m: passthrough_act}\n\u001b[0;32m---> 66\u001b[0m \u001b[43mQuantProxyMixin\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m 67\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 68\u001b[0m \u001b[43m \u001b[49m\u001b[43mquant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_quant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 69\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxy_prefix\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_proxy_prefix\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 70\u001b[0m \u001b[43m \u001b[49m\u001b[43mkwargs_prefix\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_kwargs_prefix\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 71\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxy_protocol\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mActQuantProxyProtocol\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 72\u001b[0m \u001b[43m \u001b[49m\u001b[43mnone_quant_injector\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mNoneActQuant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 73\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mprefixed_kwargs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 74\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:48\u001b[0m, in \u001b[0;36mQuantProxyMixin.__init__\u001b[0;34m(self, quant, proxy_protocol, none_quant_injector, proxy_prefix, kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 46\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant\n\u001b[1;32m 47\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant_injector\u001b[38;5;241m.\u001b[39mlet(\u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mfilter_kwargs(kwargs_prefix, kwargs))\n\u001b[0;32m---> 48\u001b[0m quant \u001b[38;5;241m=\u001b[39m \u001b[43mquant_injector\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mproxy_class\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_injector\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 49\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 50\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant, proxy_protocol):\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/runtime_quant.py:198\u001b[0m, in \u001b[0;36mActQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 197\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector):\n\u001b[0;32m--> 198\u001b[0m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mquant_layer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_injector\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 199\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcache_class \u001b[38;5;241m=\u001b[39m _CachedIO\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/runtime_quant.py:93\u001b[0m, in \u001b[0;36mActQuantProxyFromInjectorBase.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 92\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector):\n\u001b[0;32m---> 93\u001b[0m \u001b[43mQuantProxyFromInjector\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_layer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_injector\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 94\u001b[0m ActQuantProxyProtocol\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m)\n\u001b[1;32m 95\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mis_passthrough_act \u001b[38;5;241m=\u001b[39m _is_passthrough_act(quant_injector)\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/quant_proxy.py:80\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 78\u001b[0m \u001b[38;5;66;03m# Use a normal list and not a ModuleList since this is a pointer to parent modules\u001b[39;00m\n\u001b[1;32m 79\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list \u001b[38;5;241m=\u001b[39m []\n\u001b[0;32m---> 80\u001b[0m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43madd_tracked_module\u001b[49m\u001b[43m(\u001b[49m\u001b[43mquant_layer\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 81\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdisable_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m\n\u001b[1;32m 82\u001b[0m \u001b[38;5;66;03m# Torch.compile compatibility requires this\u001b[39;00m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/quant_proxy.py:120\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.add_tracked_module\u001b[0;34m(self, module)\u001b[0m\n\u001b[1;32m 118\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list\u001b[38;5;241m.\u001b[39mappend(module)\n\u001b[1;32m 119\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mupdate_tracked_modules()\n\u001b[0;32m--> 120\u001b[0m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43minit_tensor_quant\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 121\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 122\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mTrying to add None as a parent module.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/runtime_quant.py:127\u001b[0m, in \u001b[0;36mActQuantProxyFromInjectorBase.init_tensor_quant\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 126\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21minit_tensor_quant\u001b[39m(\u001b[38;5;28mself\u001b[39m):\n\u001b[0;32m--> 127\u001b[0m tensor_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mquant_injector\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtensor_quant\u001b[49m\n\u001b[1;32m 128\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector:\n\u001b[1;32m 129\u001b[0m act_impl \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector\u001b[38;5;241m.\u001b[39mact_impl\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/inject/__init__.py:129\u001b[0m, in \u001b[0;36m_ExtendedInjectorType.__getattr__\u001b[0;34m(cls, attrname)\u001b[0m\n\u001b[1;32m 126\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 127\u001b[0m message \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m can not resolve attribute \u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 128\u001b[0m \u001b[38;5;28mcls\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, current_attr)\n\u001b[0;32m--> 129\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m DependencyError(message)\n\u001b[1;32m 131\u001b[0m marker, attribute, args, have_defaults \u001b[38;5;241m=\u001b[39m spec\n\u001b[1;32m 133\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mset\u001b[39m(args)\u001b[38;5;241m.\u001b[39missubset(cached):\n", "\u001b[0;31mDependencyError\u001b[0m: 'Int8ActPerTensorFloatMinMaxInit' can not resolve attribute 'max_val' while building 'scaling_init_impl'" ] } @@ -741,6 +744,7 @@ " (delay_wrapper): DelayWrapper(\n", " (delay_impl): _NoDelay()\n", " )\n", + " (input_view_impl): Identity()\n", " )\n", " (scaling_impl): ParameterFromRuntimeStatsScaling(\n", " (stats_input_view_shape_impl): OverTensorView()\n", @@ -813,6 +817,7 @@ " (delay_wrapper): DelayWrapper(\n", " (delay_impl): _NoDelay()\n", " )\n", + " (input_view_impl): Identity()\n", " )\n", " (scaling_impl): ParameterFromRuntimeStatsScaling(\n", " (stats_input_view_shape_impl): OverOutputChannelView(\n", @@ -922,7 +927,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.10.13" } }, "nbformat": 4, diff --git a/notebooks/03_anatomy_of_a_quantizer.ipynb b/notebooks/03_anatomy_of_a_quantizer.ipynb index c4c9295e7..10de11221 100644 --- a/notebooks/03_anatomy_of_a_quantizer.ipynb +++ b/notebooks/03_anatomy_of_a_quantizer.ipynb @@ -57,6 +57,7 @@ " (delay_wrapper): DelayWrapper(\n", " (delay_impl): _NoDelay()\n", " )\n", + " (input_view_impl): Identity()\n", " )\n", " (scaling_impl): ParameterFromRuntimeStatsScaling(\n", " (stats_input_view_shape_impl): OverTensorView()\n", @@ -489,7 +490,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[ 0.1000, 0.1000, -0.1000],\n", + "IntQuantTensor(value=tensor([[[[ 0.1000, 0.1000, -0.1000],\n", " [ 0.1000, -0.1000, -0.1000],\n", " [-0.1000, -0.1000, -0.1000]],\n", "\n", @@ -563,7 +564,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[-0.1000, -0.1000, 0.1000],\n", + "IntQuantTensor(value=tensor([[[[-0.1000, -0.1000, 0.1000],\n", " [-0.1000, 0.1000, -0.1000],\n", " [ 0.1000, 0.1000, 0.1000]],\n", "\n", @@ -626,14 +627,14 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/torch/_tensor.py:1362: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449183005/work/c10/core/TensorImpl.h:1900.)\n", - " return super().rename(names)\n" + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/_tensor.py:1255: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/c10/core/TensorImpl.h:1758.)\n", + " return super(Tensor, self).rename(names)\n" ] }, { "data": { "text/plain": [ - "QuantTensor(value=tensor([[-0.1000, 0.1000, -0.1000, -0.1000],\n", + "IntQuantTensor(value=tensor([[-0.1000, 0.1000, -0.1000, -0.1000],\n", " [ 0.1000, 0.1000, 0.1000, 0.1000],\n", " [-0.1000, -0.1000, 0.1000, -0.1000],\n", " [-0.1000, 0.1000, -0.1000, 0.1000]], grad_fn=), scale=tensor(0.1000, grad_fn=), zero_point=tensor(0.), bit_width=tensor(1.), signed_t=tensor(True), training_t=tensor(True))" @@ -671,7 +672,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[ 0.0010, -0.0010, -0.0010, 0.0010],\n", + "IntQuantTensor(value=tensor([[ 0.0010, -0.0010, -0.0010, 0.0010],\n", " [ 0.0010, 0.0010, 0.0010, 0.0010],\n", " [ 0.0010, -0.0010, 0.0010, 0.0010],\n", " [ 0.0010, -0.0010, -0.0010, -0.0010]], grad_fn=), scale=tensor(0.0010, grad_fn=), zero_point=tensor(0.), bit_width=tensor(1.), signed_t=tensor(True), training_t=tensor(True))" @@ -735,7 +736,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[ 0.1820, -0.1820, -0.1820],\n", + "IntQuantTensor(value=tensor([[[[ 0.1820, -0.1820, -0.1820],\n", " [ 0.1820, 0.1820, 0.1820],\n", " [-0.1820, -0.1820, -0.1820]],\n", "\n", @@ -848,8 +849,8 @@ "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", - "Cell \u001b[0;32mIn[21], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m param_from_max_quant_conv\u001b[38;5;241m.\u001b[39mload_state_dict(float_conv\u001b[38;5;241m.\u001b[39mstate_dict())\n", - "File \u001b[0;32m/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/torch/nn/modules/module.py:2152\u001b[0m, in \u001b[0;36mModule.load_state_dict\u001b[0;34m(self, state_dict, strict, assign)\u001b[0m\n\u001b[1;32m 2147\u001b[0m error_msgs\u001b[38;5;241m.\u001b[39minsert(\n\u001b[1;32m 2148\u001b[0m \u001b[38;5;241m0\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mMissing key(s) in state_dict: \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m. \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 2149\u001b[0m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m, \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(\u001b[38;5;124mf\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{\u001b[39;00mk\u001b[38;5;132;01m}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;28;01mfor\u001b[39;00m k \u001b[38;5;129;01min\u001b[39;00m missing_keys)))\n\u001b[1;32m 2151\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(error_msgs) \u001b[38;5;241m>\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[0;32m-> 2152\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mError(s) in loading state_dict for \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m:\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 2153\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__class__\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(error_msgs)))\n\u001b[1;32m 2154\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m _IncompatibleKeys(missing_keys, unexpected_keys)\n", + "Cell \u001b[0;32mIn[21], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[43mparam_from_max_quant_conv\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mload_state_dict\u001b[49m\u001b[43m(\u001b[49m\u001b[43mfloat_conv\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mstate_dict\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1671\u001b[0m, in \u001b[0;36mModule.load_state_dict\u001b[0;34m(self, state_dict, strict)\u001b[0m\n\u001b[1;32m 1666\u001b[0m error_msgs\u001b[38;5;241m.\u001b[39minsert(\n\u001b[1;32m 1667\u001b[0m \u001b[38;5;241m0\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mMissing key(s) in state_dict: \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m. \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 1668\u001b[0m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m, \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(k) \u001b[38;5;28;01mfor\u001b[39;00m k \u001b[38;5;129;01min\u001b[39;00m missing_keys)))\n\u001b[1;32m 1670\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(error_msgs) \u001b[38;5;241m>\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[0;32m-> 1671\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mError(s) in loading state_dict for \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m:\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 1672\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__class__\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(error_msgs)))\n\u001b[1;32m 1673\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m _IncompatibleKeys(missing_keys, unexpected_keys)\n", "\u001b[0;31mRuntimeError\u001b[0m: Error(s) in loading state_dict for QuantConv2d:\n\tMissing key(s) in state_dict: \"weight_quant.tensor_quant.scaling_impl.value\". " ] } @@ -906,7 +907,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[ 0.1924, 0.1924, -0.1924],\n", + "IntQuantTensor(value=tensor([[[[ 0.1924, 0.1924, -0.1924],\n", " [ 0.1924, 0.1924, 0.1924],\n", " [ 0.1924, 0.1924, 0.1924]],\n", "\n", @@ -1261,7 +1262,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[ 0.1612, -0.1612, -0.1612],\n", + "IntQuantTensor(value=tensor([[[[ 0.1612, -0.1612, -0.1612],\n", " [-0.1612, -0.1612, -0.1612],\n", " [ 0.1612, 0.1612, 0.1612]],\n", "\n", @@ -1319,7 +1320,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[[[ 0.1924, 0.1924, -0.1924],\n", + "IntQuantTensor(value=tensor([[[[ 0.1924, 0.1924, -0.1924],\n", " [ 0.1924, 0.1924, 0.1924],\n", " [ 0.1924, 0.1924, 0.1924]],\n", "\n", @@ -1418,18 +1419,19 @@ "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mDependencyError\u001b[0m Traceback (most recent call last)", - "Cell \u001b[0;32mIn[35], line 3\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mnn\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m QuantIdentity\n\u001b[0;32m----> 3\u001b[0m quant_identity \u001b[38;5;241m=\u001b[39m QuantIdentity(\n\u001b[1;32m 4\u001b[0m act_quant\u001b[38;5;241m=\u001b[39mAdvancedActQuantizer, is_clamped\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m, scaling_per_output_channel\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/quant_activation.py:113\u001b[0m, in \u001b[0;36mQuantIdentity.__init__\u001b[0;34m(self, act_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 108\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 109\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 110\u001b[0m act_quant: Optional[ActQuantType] \u001b[38;5;241m=\u001b[39m Int8ActPerTensorFloat,\n\u001b[1;32m 111\u001b[0m return_quant_tensor: \u001b[38;5;28mbool\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m,\n\u001b[1;32m 112\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m--> 113\u001b[0m QuantNLAL\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 114\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 115\u001b[0m input_quant\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mNone\u001b[39;00m,\n\u001b[1;32m 116\u001b[0m act_impl\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mNone\u001b[39;00m,\n\u001b[1;32m 117\u001b[0m passthrough_act\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m 118\u001b[0m act_quant\u001b[38;5;241m=\u001b[39mact_quant,\n\u001b[1;32m 119\u001b[0m return_quant_tensor\u001b[38;5;241m=\u001b[39mreturn_quant_tensor,\n\u001b[1;32m 120\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/quant_layer.py:37\u001b[0m, in \u001b[0;36mQuantNonLinearActLayer.__init__\u001b[0;34m(self, act_impl, passthrough_act, input_quant, act_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 35\u001b[0m QuantLayerMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, return_quant_tensor)\n\u001b[1;32m 36\u001b[0m QuantInputMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, input_quant, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[0;32m---> 37\u001b[0m QuantNonLinearActMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, act_impl, passthrough_act, act_quant, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/mixin/act.py:118\u001b[0m, in \u001b[0;36mQuantNonLinearActMixin.__init__\u001b[0;34m(self, act_impl, passthrough_act, act_quant, act_proxy_prefix, act_kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 107\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 108\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 109\u001b[0m act_impl: Optional[Type[Module]],\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 113\u001b[0m act_kwargs_prefix\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m 114\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m 115\u001b[0m prefixed_kwargs \u001b[38;5;241m=\u001b[39m {\n\u001b[1;32m 116\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m: act_impl,\n\u001b[1;32m 117\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mpassthrough_act\u001b[39m\u001b[38;5;124m'\u001b[39m: passthrough_act}\n\u001b[0;32m--> 118\u001b[0m QuantProxyMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 119\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 120\u001b[0m quant\u001b[38;5;241m=\u001b[39mact_quant,\n\u001b[1;32m 121\u001b[0m proxy_prefix\u001b[38;5;241m=\u001b[39mact_proxy_prefix,\n\u001b[1;32m 122\u001b[0m kwargs_prefix\u001b[38;5;241m=\u001b[39mact_kwargs_prefix,\n\u001b[1;32m 123\u001b[0m proxy_protocol\u001b[38;5;241m=\u001b[39mActQuantProxyProtocol,\n\u001b[1;32m 124\u001b[0m none_quant_injector\u001b[38;5;241m=\u001b[39mNoneActQuant,\n\u001b[1;32m 125\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mprefixed_kwargs,\n\u001b[1;32m 126\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:71\u001b[0m, in \u001b[0;36mQuantProxyMixin.__init__\u001b[0;34m(self, quant, proxy_protocol, none_quant_injector, proxy_prefix, kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 69\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant\n\u001b[1;32m 70\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant_injector\u001b[38;5;241m.\u001b[39mlet(\u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mfilter_kwargs(kwargs_prefix, kwargs))\n\u001b[0;32m---> 71\u001b[0m quant \u001b[38;5;241m=\u001b[39m quant_injector\u001b[38;5;241m.\u001b[39mproxy_class(\u001b[38;5;28mself\u001b[39m, quant_injector)\n\u001b[1;32m 72\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 73\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant, proxy_protocol):\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/runtime_quant.py:89\u001b[0m, in \u001b[0;36mActQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 88\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector):\n\u001b[0;32m---> 89\u001b[0m QuantProxyFromInjector\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector)\n\u001b[1;32m 90\u001b[0m ActQuantProxyProtocol\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m)\n\u001b[1;32m 91\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mis_passthrough_act \u001b[38;5;241m=\u001b[39m _is_passthrough_act(quant_injector)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/quant_proxy.py:89\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 87\u001b[0m \u001b[38;5;66;03m# Use a normal list and not a ModuleList since this is a pointer to parent modules\u001b[39;00m\n\u001b[1;32m 88\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list \u001b[38;5;241m=\u001b[39m []\n\u001b[0;32m---> 89\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39madd_tracked_module(quant_layer)\n\u001b[1;32m 90\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdisable_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/quant_proxy.py:131\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.add_tracked_module\u001b[0;34m(self, module)\u001b[0m\n\u001b[1;32m 129\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list\u001b[38;5;241m.\u001b[39mappend(module)\n\u001b[1;32m 130\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mupdate_tracked_modules()\n\u001b[0;32m--> 131\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39minit_tensor_quant()\n\u001b[1;32m 132\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 133\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mTrying to add None as a parent module.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/proxy/runtime_quant.py:102\u001b[0m, in \u001b[0;36mActQuantProxyFromInjector.init_tensor_quant\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 101\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21minit_tensor_quant\u001b[39m(\u001b[38;5;28mself\u001b[39m):\n\u001b[0;32m--> 102\u001b[0m tensor_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector\u001b[38;5;241m.\u001b[39mtensor_quant\n\u001b[1;32m 103\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector:\n\u001b[1;32m 104\u001b[0m act_impl \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector\u001b[38;5;241m.\u001b[39mact_impl\n", + "Cell \u001b[0;32mIn[35], line 3\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mbrevitas\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mnn\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m QuantIdentity\n\u001b[0;32m----> 3\u001b[0m quant_identity \u001b[38;5;241m=\u001b[39m \u001b[43mQuantIdentity\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 4\u001b[0m \u001b[43m \u001b[49m\u001b[43mact_quant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mAdvancedActQuantizer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mis_clamped\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mscaling_per_output_channel\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_activation.py:113\u001b[0m, in \u001b[0;36mQuantIdentity.__init__\u001b[0;34m(self, act_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 108\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 109\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 110\u001b[0m act_quant: Optional[ActQuantType] \u001b[38;5;241m=\u001b[39m Int8ActPerTensorFloat,\n\u001b[1;32m 111\u001b[0m return_quant_tensor: \u001b[38;5;28mbool\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m,\n\u001b[1;32m 112\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m--> 113\u001b[0m \u001b[43mQuantNLAL\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m 114\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 115\u001b[0m \u001b[43m \u001b[49m\u001b[43minput_quant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 116\u001b[0m \u001b[43m \u001b[49m\u001b[43mact_impl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 117\u001b[0m \u001b[43m \u001b[49m\u001b[43mpassthrough_act\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 118\u001b[0m \u001b[43m \u001b[49m\u001b[43mact_quant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_quant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 119\u001b[0m \u001b[43m \u001b[49m\u001b[43mreturn_quant_tensor\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mreturn_quant_tensor\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 120\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_layer.py:34\u001b[0m, in \u001b[0;36mQuantNonLinearActLayer.__init__\u001b[0;34m(self, act_impl, passthrough_act, input_quant, act_quant, return_quant_tensor, **kwargs)\u001b[0m\n\u001b[1;32m 32\u001b[0m QuantLayerMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, return_quant_tensor)\n\u001b[1;32m 33\u001b[0m QuantInputMixin\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, input_quant, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[0;32m---> 34\u001b[0m \u001b[43mQuantNonLinearActMixin\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mact_impl\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mpassthrough_act\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mact_quant\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/act.py:66\u001b[0m, in \u001b[0;36mQuantNonLinearActMixin.__init__\u001b[0;34m(self, act_impl, passthrough_act, act_quant, act_proxy_prefix, act_kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 55\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\n\u001b[1;32m 56\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 57\u001b[0m act_impl: Optional[Type[Module]],\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 61\u001b[0m act_kwargs_prefix\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m 62\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m 63\u001b[0m prefixed_kwargs \u001b[38;5;241m=\u001b[39m {\n\u001b[1;32m 64\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m: act_impl,\n\u001b[1;32m 65\u001b[0m act_kwargs_prefix \u001b[38;5;241m+\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mpassthrough_act\u001b[39m\u001b[38;5;124m'\u001b[39m: passthrough_act}\n\u001b[0;32m---> 66\u001b[0m \u001b[43mQuantProxyMixin\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m 67\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 68\u001b[0m \u001b[43m \u001b[49m\u001b[43mquant\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_quant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 69\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxy_prefix\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_proxy_prefix\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 70\u001b[0m \u001b[43m \u001b[49m\u001b[43mkwargs_prefix\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mact_kwargs_prefix\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 71\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxy_protocol\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mActQuantProxyProtocol\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 72\u001b[0m \u001b[43m \u001b[49m\u001b[43mnone_quant_injector\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mNoneActQuant\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 73\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mprefixed_kwargs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 74\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:48\u001b[0m, in \u001b[0;36mQuantProxyMixin.__init__\u001b[0;34m(self, quant, proxy_protocol, none_quant_injector, proxy_prefix, kwargs_prefix, **kwargs)\u001b[0m\n\u001b[1;32m 46\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant\n\u001b[1;32m 47\u001b[0m quant_injector \u001b[38;5;241m=\u001b[39m quant_injector\u001b[38;5;241m.\u001b[39mlet(\u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mfilter_kwargs(kwargs_prefix, kwargs))\n\u001b[0;32m---> 48\u001b[0m quant \u001b[38;5;241m=\u001b[39m \u001b[43mquant_injector\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mproxy_class\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_injector\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 49\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 50\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(quant, proxy_protocol):\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/runtime_quant.py:198\u001b[0m, in \u001b[0;36mActQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 197\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector):\n\u001b[0;32m--> 198\u001b[0m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mquant_layer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_injector\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 199\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcache_class \u001b[38;5;241m=\u001b[39m _CachedIO\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/runtime_quant.py:93\u001b[0m, in \u001b[0;36mActQuantProxyFromInjectorBase.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 92\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, quant_layer, quant_injector):\n\u001b[0;32m---> 93\u001b[0m \u001b[43mQuantProxyFromInjector\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_layer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mquant_injector\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 94\u001b[0m ActQuantProxyProtocol\u001b[38;5;241m.\u001b[39m\u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m)\n\u001b[1;32m 95\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mis_passthrough_act \u001b[38;5;241m=\u001b[39m _is_passthrough_act(quant_injector)\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/quant_proxy.py:80\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.__init__\u001b[0;34m(self, quant_layer, quant_injector)\u001b[0m\n\u001b[1;32m 78\u001b[0m \u001b[38;5;66;03m# Use a normal list and not a ModuleList since this is a pointer to parent modules\u001b[39;00m\n\u001b[1;32m 79\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list \u001b[38;5;241m=\u001b[39m []\n\u001b[0;32m---> 80\u001b[0m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43madd_tracked_module\u001b[49m\u001b[43m(\u001b[49m\u001b[43mquant_layer\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 81\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdisable_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m\n\u001b[1;32m 82\u001b[0m \u001b[38;5;66;03m# Torch.compile compatibility requires this\u001b[39;00m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/quant_proxy.py:120\u001b[0m, in \u001b[0;36mQuantProxyFromInjector.add_tracked_module\u001b[0;34m(self, module)\u001b[0m\n\u001b[1;32m 118\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtracked_module_list\u001b[38;5;241m.\u001b[39mappend(module)\n\u001b[1;32m 119\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mupdate_tracked_modules()\n\u001b[0;32m--> 120\u001b[0m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43minit_tensor_quant\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 121\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 122\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mTrying to add None as a parent module.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/proxy/runtime_quant.py:127\u001b[0m, in \u001b[0;36mActQuantProxyFromInjectorBase.init_tensor_quant\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 126\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21minit_tensor_quant\u001b[39m(\u001b[38;5;28mself\u001b[39m):\n\u001b[0;32m--> 127\u001b[0m tensor_quant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mquant_injector\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtensor_quant\u001b[49m\n\u001b[1;32m 128\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mact_impl\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector:\n\u001b[1;32m 129\u001b[0m act_impl \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mquant_injector\u001b[38;5;241m.\u001b[39mact_impl\n", " \u001b[0;31m[... skipping hidden 1 frame]\u001b[0m\n", - "File \u001b[0;32m/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/_dependencies/this.py:51\u001b[0m, in \u001b[0;36m_ThisSpec.__call__\u001b[0;34m(self, __self__)\u001b[0m\n\u001b[1;32m 49\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m kind \u001b[38;5;241m==\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m.\u001b[39m\u001b[38;5;124m\"\u001b[39m:\n\u001b[1;32m 50\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m---> 51\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mgetattr\u001b[39m(result, symbol)\n\u001b[1;32m 52\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m DependencyError:\n\u001b[1;32m 53\u001b[0m message \u001b[38;5;241m=\u001b[39m (\n\u001b[1;32m 54\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mYou tried to shift this more times than Injector has levels\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 55\u001b[0m )\n", - "File \u001b[0;32m/scratch/fabian/brevitas/src/brevitas/inject/__init__.py:129\u001b[0m, in \u001b[0;36m_ExtendedInjectorType.__getattr__\u001b[0;34m(cls, attrname)\u001b[0m\n\u001b[1;32m 126\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 127\u001b[0m message \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m can not resolve attribute \u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 128\u001b[0m \u001b[38;5;28mcls\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, current_attr)\n\u001b[0;32m--> 129\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m DependencyError(message)\n\u001b[1;32m 131\u001b[0m marker, attribute, args, have_defaults \u001b[38;5;241m=\u001b[39m spec\n\u001b[1;32m 133\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mset\u001b[39m(args)\u001b[38;5;241m.\u001b[39missubset(cached):\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/_dependencies/this.py:51\u001b[0m, in \u001b[0;36m_ThisSpec.__call__\u001b[0;34m(self, __self__)\u001b[0m\n\u001b[1;32m 49\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m kind \u001b[38;5;241m==\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m.\u001b[39m\u001b[38;5;124m\"\u001b[39m:\n\u001b[1;32m 50\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m---> 51\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mgetattr\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mresult\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msymbol\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 52\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m DependencyError:\n\u001b[1;32m 53\u001b[0m message \u001b[38;5;241m=\u001b[39m (\n\u001b[1;32m 54\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mYou tried to shift this more times than Injector has levels\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 55\u001b[0m )\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/inject/__init__.py:129\u001b[0m, in \u001b[0;36m_ExtendedInjectorType.__getattr__\u001b[0;34m(cls, attrname)\u001b[0m\n\u001b[1;32m 126\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 127\u001b[0m message \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m can not resolve attribute \u001b[39m\u001b[38;5;132;01m{!r}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 128\u001b[0m \u001b[38;5;28mcls\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, current_attr)\n\u001b[0;32m--> 129\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m DependencyError(message)\n\u001b[1;32m 131\u001b[0m marker, attribute, args, have_defaults \u001b[38;5;241m=\u001b[39m spec\n\u001b[1;32m 133\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mset\u001b[39m(args)\u001b[38;5;241m.\u001b[39missubset(cached):\n", "\u001b[0;31mDependencyError\u001b[0m: 'AdvancedActQuantizer' can not resolve attribute 'per_channel_broadcastable_shape'" ] } @@ -1456,7 +1458,7 @@ { "data": { "text/plain": [ - "QuantTensor(value=tensor([[ 0.0100, 0.0100, 0.0100, -0.0100],\n", + "IntQuantTensor(value=tensor([[ 0.0100, 0.0100, 0.0100, -0.0100],\n", " [ 0.0100, -0.0100, -0.0100, 0.0100],\n", " [-0.0100, -0.0100, -0.0100, 0.0100],\n", " [ 0.0100, 0.0100, 0.0100, 0.0100]], grad_fn=), scale=tensor([[0.0100],\n", @@ -1501,7 +1503,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.10.13" } }, "nbformat": 4, diff --git a/notebooks/Brevitas_TVMCon2021.ipynb b/notebooks/Brevitas_TVMCon2021.ipynb index 7f5846e09..00fbd6c0c 100644 --- a/notebooks/Brevitas_TVMCon2021.ipynb +++ b/notebooks/Brevitas_TVMCon2021.ipynb @@ -208,9 +208,9 @@ "name": "stderr", "output_type": "stream", "text": [ - "/home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/_tensor.py:1419: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/c10/core/TensorImpl.h:1921.)\n", - " return super().rename(names)\n", - "/home/giuseppe/Documents/git/dev_brevitas/src/brevitas/nn/quant_linear.py:69: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/_tensor.py:1255: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/c10/core/TensorImpl.h:1758.)\n", + " return super(Tensor, self).rename(names)\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_linear.py:69: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " output_tensor = linear(x, quant_weight, quant_bias)\n" ] } @@ -634,10 +634,9 @@ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", "Cell \u001b[0;32mIn[13], line 8\u001b[0m\n\u001b[1;32m 5\u001b[0m float_input \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39mrandn(\u001b[38;5;241m3\u001b[39m, \u001b[38;5;241m2\u001b[39m)\n\u001b[1;32m 6\u001b[0m quant_linear \u001b[38;5;241m=\u001b[39m QuantLinear(\u001b[38;5;241m2\u001b[39m, \u001b[38;5;241m4\u001b[39m, bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m, bias_quant\u001b[38;5;241m=\u001b[39mInt16Bias, return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m----> 8\u001b[0m quant_output \u001b[38;5;241m=\u001b[39m \u001b[43mquant_linear\u001b[49m\u001b[43m(\u001b[49m\u001b[43mfloat_input\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/nn/modules/module.py:1532\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1530\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1531\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1532\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/nn/modules/module.py:1541\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1536\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1537\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1538\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1539\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1540\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1541\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1543\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 1544\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n", - "File \u001b[0;32m~/Documents/git/dev_brevitas/src/brevitas/nn/quant_linear.py:66\u001b[0m, in \u001b[0;36mQuantLinear.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 65\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[0;32m---> 66\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/Documents/git/dev_brevitas/src/brevitas/nn/quant_layer.py:152\u001b[0m, in \u001b[0;36mQuantWeightBiasInputOutputLayer.forward_impl\u001b[0;34m(self, inp)\u001b[0m\n\u001b[1;32m 148\u001b[0m compute_output_quant_tensor \u001b[38;5;241m=\u001b[39m \u001b[38;5;28misinstance\u001b[39m(quant_input, QuantTensor) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(\n\u001b[1;32m 149\u001b[0m quant_weight, QuantTensor)\n\u001b[1;32m 150\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (compute_output_quant_tensor \u001b[38;5;129;01mor\u001b[39;00m\n\u001b[1;32m 151\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39moutput_quant\u001b[38;5;241m.\u001b[39mis_quant_enabled) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreturn_quant_tensor:\n\u001b[0;32m--> 152\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mQuantLayer is not correctly configured\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 154\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[1;32m 155\u001b[0m quant_bias \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias_quant(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias, quant_input, quant_weight)\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1194\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m 1190\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1191\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1192\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1193\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1194\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1195\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m 1196\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_linear.py:66\u001b[0m, in \u001b[0;36mQuantLinear.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 65\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Union[Tensor, QuantTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Union[Tensor, QuantTensor]:\n\u001b[0;32m---> 66\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mforward_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_layer.py:152\u001b[0m, in \u001b[0;36mQuantWeightBiasInputOutputLayer.forward_impl\u001b[0;34m(self, inp)\u001b[0m\n\u001b[1;32m 148\u001b[0m compute_output_quant_tensor \u001b[38;5;241m=\u001b[39m \u001b[38;5;28misinstance\u001b[39m(quant_input, QuantTensor) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(\n\u001b[1;32m 149\u001b[0m quant_weight, QuantTensor)\n\u001b[1;32m 150\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (compute_output_quant_tensor \u001b[38;5;129;01mor\u001b[39;00m\n\u001b[1;32m 151\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39moutput_quant\u001b[38;5;241m.\u001b[39mis_quant_enabled) \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreturn_quant_tensor:\n\u001b[0;32m--> 152\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mQuantLayer is not correctly configured\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 154\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[1;32m 155\u001b[0m quant_bias \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias_quant(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mbias, quant_input, quant_weight)\n", "\u001b[0;31mRuntimeError\u001b[0m: QuantLayer is not correctly configured" ] } @@ -870,7 +869,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/tmp/ipykernel_751241/661358273.py:7: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/tmp/ipykernel_1328/661358273.py:7: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " quant_output = torch.tanh(quant_input)\n" ] } @@ -923,9 +922,9 @@ "name": "stderr", "output_type": "stream", "text": [ - "/tmp/ipykernel_751241/3932472163.py:8: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/tmp/ipykernel_1328/3932472163.py:8: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " train_mode_cat = torch.cat([quant_identity(float_inp1), quant_identity(float_inp2)], dim=1)\n", - "/tmp/ipykernel_751241/3932472163.py:14: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/tmp/ipykernel_1328/3932472163.py:14: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " eval_mode_cat = torch.cat([eval_quant_inp1, eval_quant_inp2], dim=1)\n" ] } @@ -1014,10 +1013,13 @@ "output_type": "stream", "text": [ "Weight QuantTensor:\n", - " IntQuantTensor(value=tensor([[-0.0000, 0.3880],\n", + " IntQuantTensor(value=tensor([[-0.0000, 0.3793],\n", " [-0.5820, -0.5044],\n", - " [-0.2716, 0.1940],\n", - " [-0.0000, 0.5432]], grad_fn=), scale=0.03879871591925621, zero_point=0.0, bit_width=5.0, signed_t=True, training_t=True)\n" + " [-0.2723, 0.1816],\n", + " [-0.0000, 0.5607]], grad_fn=), scale=tensor([[0.0253],\n", + " [0.0388],\n", + " [0.0182],\n", + " [0.0374]], grad_fn=), zero_point=0.0, bit_width=5.0, signed_t=True, training_t=True)\n" ] } ], @@ -1123,14 +1125,16 @@ "\n", "Per-channel quant output:\n", " IntQuantTensor(value=tensor([[[ 0.8616, -0.7012, 0.4503],\n", - " [-1.2235, -0.4359, -0.0473]]], grad_fn=), scale=tensor([[[0.0021]]], grad_fn=), zero_point=tensor([0.]), bit_width=17.0, signed_t=True, training_t=True)\n" + " [-1.1285, -0.4937, -0.1901]]], grad_fn=), scale=tensor([[[0.0021],\n", + " [0.0013]]], grad_fn=), zero_point=tensor([[[-254.0000],\n", + " [ 406.0000]]], grad_fn=), bit_width=17.0, signed_t=True, training_t=True)\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "/home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/nn/modules/conv.py:306: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/conv.py:309: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " return F.conv1d(input, weight, bias, self.stride,\n" ] } @@ -1188,7 +1192,9 @@ "\n", "Per-channel quant output:\n", " IntQuantTensor(value=tensor([[[ 0.8616, -0.7012, 0.4503],\n", - " [-1.2235, -0.4359, -0.0473]]], grad_fn=), scale=tensor([[[0.0021]]], grad_fn=), zero_point=tensor([0.]), bit_width=17.0, signed_t=True, training_t=True)\n" + " [-1.1285, -0.4937, -0.1901]]], grad_fn=), scale=tensor([[[0.0021],\n", + " [0.0013]]], grad_fn=), zero_point=tensor([[[-254.0000],\n", + " [ 406.0000]]], grad_fn=), bit_width=17.0, signed_t=True, training_t=True)\n" ] } ], @@ -1438,7 +1444,7 @@ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", "Cell \u001b[0;32mIn[29], line 12\u001b[0m\n\u001b[1;32m 5\u001b[0m float_linear \u001b[38;5;241m=\u001b[39m nn\u001b[38;5;241m.\u001b[39mLinear(\u001b[38;5;241m2\u001b[39m, \u001b[38;5;241m4\u001b[39m, bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m)\n\u001b[1;32m 6\u001b[0m quant_linear \u001b[38;5;241m=\u001b[39m QuantLinear(\n\u001b[1;32m 7\u001b[0m \u001b[38;5;241m2\u001b[39m, \u001b[38;5;241m4\u001b[39m, \n\u001b[1;32m 8\u001b[0m input_quant\u001b[38;5;241m=\u001b[39mLearnedIntActPerTensorFloat,\n\u001b[1;32m 9\u001b[0m weight_quant\u001b[38;5;241m=\u001b[39mLearnedIntWeightPerChannelFloat, \n\u001b[1;32m 10\u001b[0m return_quant_tensor\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m, bias\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m)\n\u001b[0;32m---> 12\u001b[0m \u001b[43mquant_linear\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mload_state_dict\u001b[49m\u001b[43m(\u001b[49m\u001b[43mfloat_linear\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mstate_dict\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", - "File \u001b[0;32m~/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/nn/modules/module.py:2189\u001b[0m, in \u001b[0;36mModule.load_state_dict\u001b[0;34m(self, state_dict, strict, assign)\u001b[0m\n\u001b[1;32m 2184\u001b[0m error_msgs\u001b[38;5;241m.\u001b[39minsert(\n\u001b[1;32m 2185\u001b[0m \u001b[38;5;241m0\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mMissing key(s) in state_dict: \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m. \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 2186\u001b[0m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m, \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(\u001b[38;5;124mf\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{\u001b[39;00mk\u001b[38;5;132;01m}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;28;01mfor\u001b[39;00m k \u001b[38;5;129;01min\u001b[39;00m missing_keys)))\n\u001b[1;32m 2188\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(error_msgs) \u001b[38;5;241m>\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[0;32m-> 2189\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mError(s) in loading state_dict for \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m:\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 2190\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__class__\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(error_msgs)))\n\u001b[1;32m 2191\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m _IncompatibleKeys(missing_keys, unexpected_keys)\n", + "File \u001b[0;32m/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/module.py:1671\u001b[0m, in \u001b[0;36mModule.load_state_dict\u001b[0;34m(self, state_dict, strict)\u001b[0m\n\u001b[1;32m 1666\u001b[0m error_msgs\u001b[38;5;241m.\u001b[39minsert(\n\u001b[1;32m 1667\u001b[0m \u001b[38;5;241m0\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mMissing key(s) in state_dict: \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m. \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 1668\u001b[0m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124m, \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(k) \u001b[38;5;28;01mfor\u001b[39;00m k \u001b[38;5;129;01min\u001b[39;00m missing_keys)))\n\u001b[1;32m 1670\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(error_msgs) \u001b[38;5;241m>\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[0;32m-> 1671\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mError(s) in loading state_dict for \u001b[39m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m:\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;132;01m{}\u001b[39;00m\u001b[38;5;124m'\u001b[39m\u001b[38;5;241m.\u001b[39mformat(\n\u001b[1;32m 1672\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__class__\u001b[39m\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;241m.\u001b[39mjoin(error_msgs)))\n\u001b[1;32m 1673\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m _IncompatibleKeys(missing_keys, unexpected_keys)\n", "\u001b[0;31mRuntimeError\u001b[0m: Error(s) in loading state_dict for QuantLinear:\n\tMissing key(s) in state_dict: \"input_quant.fused_activation_quant_proxy.tensor_quant.scaling_impl.value\", \"input_quant.fused_activation_quant_proxy.tensor_quant.msb_clamp_bit_width_impl.bit_width_offset\", \"weight_quant.tensor_quant.scaling_impl.value\", \"weight_quant.tensor_quant.msb_clamp_bit_width_impl.bit_width_offset\". " ] } @@ -1598,6 +1604,7 @@ " (delay_wrapper): DelayWrapper(\n", " (delay_impl): _NoDelay()\n", " )\n", + " (input_view_impl): Identity()\n", " )\n", " (scaling_impl): ParameterFromRuntimeStatsScaling(\n", " (stats_input_view_shape_impl): OverTensorView()\n", @@ -1884,12 +1891,11 @@ "name": "stdout", "output_type": "stream", "text": [ - "\u001b[33mDEPRECATION: Loading egg at /home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torchao-0.3.0-py3.11.egg is deprecated. pip 24.3 will enforce this behaviour change. A possible replacement is to use pip for package installation.. Discussion can be found at https://github.com/pypa/pip/issues/12330\u001b[0m\u001b[33m\n", - "\u001b[0mRequirement already satisfied: netron in /home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages (7.6.3)\n", - "Requirement already satisfied: onnx in /home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages (1.15.0)\n", - "Requirement already satisfied: onnxoptimizer in /home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages (0.3.13)\n", - "Requirement already satisfied: numpy in /home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages (from onnx) (1.26.4)\n", - "Requirement already satisfied: protobuf>=3.20.2 in /home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages (from onnx) (3.20.3)\n" + "Requirement already satisfied: netron in /proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages (7.2.9)\r\n", + "Requirement already satisfied: onnx in /proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages (1.15.0)\r\n", + "Requirement already satisfied: onnxoptimizer in /proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages (0.3.13)\r\n", + "Requirement already satisfied: numpy in /proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages (from onnx) (1.26.0)\r\n", + "Requirement already satisfied: protobuf>=3.20.2 in /proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages (from onnx) (4.24.4)\r\n" ] } ], @@ -1931,8 +1937,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", - " from .autonotebook import tqdm as notebook_tqdm\n" + "[W NNPACK.cpp:53] Could not initialize NNPACK! Reason: Unsupported hardware.\n" ] } ], @@ -2147,7 +2152,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.9" + "version": "3.10.13" }, "vscode": { "interpreter": { diff --git a/notebooks/ONNX_export_tutorial.ipynb b/notebooks/ONNX_export_tutorial.ipynb index 65ef2ea58..a315178af 100644 --- a/notebooks/ONNX_export_tutorial.ipynb +++ b/notebooks/ONNX_export_tutorial.ipynb @@ -32,7 +32,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "Requirement already satisfied: netron in /scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages (7.4.5)\n", + "Requirement already satisfied: netron in /proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages (7.2.9)\r\n", "Note: you may need to restart the kernel to use updated packages.\n" ] } @@ -555,7 +555,9 @@ "name": "stderr", "output_type": "stream", "text": [ - "2024-03-06 02:12:47.492497092 [W:onnxruntime:, graph.cc:1283 Graph] Initializer linear.bias appears in graph inputs and will not be treated as constant value/weight. This may prevent some of the graph optimizations, like const folding. Move it out of graph inputs if there is no need to override it, by either re-generating the model with latest exporter/converter or with the tool onnxruntime/tools/python/remove_initializer_from_input.py.\n" + "2024-09-12 12:18:03.405472924 [W:onnxruntime:, graph.cc:1283 Graph] Initializer linear.bias appears in graph inputs and will not be treated as constant value/weight. This may prevent some of the graph optimizations, like const folding. Move it out of graph inputs if there is no need to override it, by either re-generating the model with latest exporter/converter or with the tool onnxruntime/tools/python/remove_initializer_from_input.py.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/quant_linear.py:69: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", + " output_tensor = linear(x, quant_weight, quant_bias)\n" ] } ], @@ -771,7 +773,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.10.13" }, "vscode": { "interpreter": { diff --git a/notebooks/minifloat_mx_tutorial.ipynb b/notebooks/minifloat_mx_tutorial.ipynb index bd43880de..2a6f9bccb 100644 --- a/notebooks/minifloat_mx_tutorial.ipynb +++ b/notebooks/minifloat_mx_tutorial.ipynb @@ -67,7 +67,10 @@ "name": "stderr", "output_type": "stream", "text": [ - "/home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/nn/modules/conv.py:456: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/_tensor.py:1255: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/c10/core/TensorImpl.h:1758.)\n", + " return super(Tensor, self).rename(names)\n", + "[W NNPACK.cpp:53] Could not initialize NNPACK! Reason: Unsupported hardware.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/nn/modules/conv.py:459: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " return F.conv2d(input, weight, bias, self.stride,\n" ] } @@ -152,7 +155,16 @@ "cell_type": "code", "execution_count": 2, "metadata": {}, - "outputs": [], + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/quant/solver/act.py:132: UserWarning: Group dim is being selected assuming batched input. Using unbatched input will fail and requires manually specification of group_dim\n", + " warn(\n" + ] + } + ], "source": [ "from brevitas.quant_tensor import GroupwiseFloatQuantTensor\n", "\n", @@ -204,14 +216,6 @@ "Non padding weights shape torch.Size([64, 8, 3, 3])\n", "Padded weights shape torch.Size([64, 32, 3, 3])\n" ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "/home/giuseppe/miniconda3/envs/brevitas_dev/lib/python3.11/site-packages/torch/nn/modules/conv.py:456: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1712608853099/work/torch/csrc/utils/python_arg_parser.cpp:294.)\n", - " return F.conv2d(input, weight, bias, self.stride,\n" - ] } ], "source": [ @@ -353,7 +357,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.9" + "version": "3.10.13" } }, "nbformat": 4, diff --git a/notebooks/quantized_recurrent.ipynb b/notebooks/quantized_recurrent.ipynb index 9700eb177..cd13fbf79 100644 --- a/notebooks/quantized_recurrent.ipynb +++ b/notebooks/quantized_recurrent.ipynb @@ -116,7 +116,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:78: UserWarning: Keyword arguments are being passed but they not being used.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:55: UserWarning: Keyword arguments are being passed but they not being used.\n", " warn('Keyword arguments are being passed but they not being used.')\n" ] } @@ -384,14 +384,14 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:307: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449183005/work/torch/csrc/utils/python_arg_parser.cpp:368.)\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:216: UserWarning: Defining your `__torch_function__` as a plain method is deprecated and will be an error in future, please define it as a classmethod. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/torch/csrc/utils/python_arg_parser.cpp:350.)\n", " return torch.cat(outputs, dim=seq_dim)\n" ] }, { "data": { "text/plain": [ - "(QuantTensor(value=tensor([[[-0.0062, -0.2872, 0.7931, 0.4309, 0.5495, -0.4558, 0.2373,\n", + "(IntQuantTensor(value=tensor([[[-0.0062, -0.2872, 0.7931, 0.4309, 0.5495, -0.4558, 0.2373,\n", " 0.6807, 0.4621, 0.6120, -0.1124, 0.3872, 0.3060, 0.7681,\n", " -0.3684, 0.0437, -0.7369, -0.3247, 0.7743, 0.3372],\n", " [ 0.5450, 0.2962, -0.3969, 0.3555, -0.5628, 0.2429, -0.4976,\n", @@ -423,7 +423,7 @@ " -0.2664, 0.4923, 0.2143, -0.4170, 0.4112, 0.5502, 0.7066,\n", " -0.6024, 0.7356, 0.0348, 0.1043, -0.1911, -0.4518]]],\n", " grad_fn=), scale=tensor(0.0059, grad_fn=), zero_point=tensor(0.), bit_width=tensor(8.), signed_t=tensor(True), training_t=tensor(True)),\n", - " QuantTensor(value=tensor([[[ 0.0579, -0.0058, -0.4054, -0.1564, -0.5560, -0.3301, 0.3533,\n", + " IntQuantTensor(value=tensor([[[ 0.0579, -0.0058, -0.4054, -0.1564, -0.5560, -0.3301, 0.3533,\n", " 0.0058, -0.1622, -0.3765, 0.1216, 0.0695, -0.4054, 0.0927,\n", " 0.6139, -0.1390, 0.7066, 0.1274, 0.1622, -0.2896],\n", " [-0.0290, -0.1738, 0.0695, 0.3765, 0.1738, 0.0579, -0.4054,\n", @@ -462,14 +462,14 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/miniforge3/envs/torchgpu/lib/python3.11/site-packages/torch/_tensor.py:1362: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449183005/work/c10/core/TensorImpl.h:1900.)\n", - " return super().rename(names)\n" + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/torch/_tensor.py:1255: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /opt/conda/conda-bld/pytorch_1670525541990/work/c10/core/TensorImpl.h:1758.)\n", + " return super(Tensor, self).rename(names)\n" ] }, { "data": { "text/plain": [ - "(QuantTensor(value=tensor([[[ 0.2111, 0.1267, 0.0060, 0.6153, -0.7721, -0.3740, -0.5188,\n", + "(IntQuantTensor(value=tensor([[[ 0.2111, 0.1267, 0.0060, 0.6153, -0.7721, -0.3740, -0.5188,\n", " 0.6273, 0.4162, 0.2051, 0.2292, 0.7239, 0.6032, 0.2533,\n", " 0.5067, 0.6635, 0.1206, -0.5730, 0.0483, 0.3318],\n", " [ 0.5742, 0.0194, -0.3807, -0.0710, -0.6000, 0.1807, 0.1355,\n", @@ -501,7 +501,7 @@ " 0.4136, 0.5383, -0.3085, 0.4070, 0.4070, 0.6630, -0.0263,\n", " 0.2823, -0.1510, 0.1313, -0.5186, 0.4464, -0.0066]]],\n", " grad_fn=), scale=tensor(0.0062, grad_fn=), zero_point=tensor(0.), bit_width=tensor(8.), signed_t=tensor(True), training_t=tensor(True)),\n", - " QuantTensor(value=tensor([[[ 0.0000, -0.4004, 0.3151, -0.0263, -0.5842, -0.1641, -0.3939,\n", + " IntQuantTensor(value=tensor([[[ 0.0000, -0.4004, 0.3151, -0.0263, -0.5842, -0.1641, -0.3939,\n", " 0.0263, -0.2429, 0.6499, -0.5186, 0.1247, -0.2101, 0.8337,\n", " -0.1444, 0.6762, -0.1641, -0.5317, -0.1707, -0.0197],\n", " [ 0.3479, 0.5974, -0.3939, 0.1444, -0.6762, 0.1969, -0.6499,\n", @@ -538,7 +538,7 @@ { "data": { "text/plain": [ - "(QuantTensor(value=tensor([[[-0.3777, -0.2074, 0.7184, 0.9110, 0.0148, -0.1926, -0.7110,\n", + "(IntQuantTensor(value=tensor([[[-0.3777, -0.2074, 0.7184, 0.9110, 0.0148, -0.1926, -0.7110,\n", " 0.1926, -0.4222, -0.9480, 0.2592, 0.2222, -0.2370, -0.5407,\n", " 0.5851, -0.2370, 0.3555, 0.1703, 0.4444, -0.2222],\n", " [ 0.4814, -0.7355, -0.1605, 0.3878, -0.5282, 0.2073, 0.0000,\n", @@ -570,7 +570,7 @@ " -0.3422, 0.8028, 0.0855, -0.7238, -0.6317, 0.2764, -0.0461,\n", " -0.4211, -0.5988, 0.2632, 0.4014, -0.7501, -0.5659]]],\n", " grad_fn=), scale=tensor(0.0069, grad_fn=), zero_point=tensor(0.), bit_width=tensor(8.), signed_t=tensor(True), training_t=tensor(True)),\n", - " QuantTensor(value=tensor([[[-0.0066, 0.4804, 0.0066, -0.1184, 0.6843, -0.0197, 0.1448,\n", + " IntQuantTensor(value=tensor([[[-0.0066, 0.4804, 0.0066, -0.1184, 0.6843, -0.0197, 0.1448,\n", " 0.1842, 0.6383, -0.1908, -0.0066, -0.1053, -0.1316, 0.0461,\n", " -0.0066, -0.2764, 0.3751, 0.3619, 0.5001, -0.1316],\n", " [-0.6449, 0.5856, -0.0263, -0.0197, 0.8357, -0.5856, 0.0395,\n", @@ -985,7 +985,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "2024-02-28 05:30:11.212034979 [W:onnxruntime:, graph.cc:1283 Graph] Initializer onnx::LSTM_93 appears in graph inputs and will not be treated as constant value/weight. This may prevent some of the graph optimizations, like const folding. Move it out of graph inputs if there is no need to override it, by either re-generating the model with latest exporter/converter or with the tool onnxruntime/tools/python/remove_initializer_from_input.py.\n" + "2024-09-12 12:18:52.692518968 [W:onnxruntime:, graph.cc:1283 Graph] Initializer onnx::LSTM_93 appears in graph inputs and will not be treated as constant value/weight. This may prevent some of the graph optimizations, like const folding. Move it out of graph inputs if there is no need to override it, by either re-generating the model with latest exporter/converter or with the tool onnxruntime/tools/python/remove_initializer_from_input.py.\n" ] } ], @@ -1084,7 +1084,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "2024-02-28 05:30:18.857576114 [W:onnxruntime:, graph.cc:1283 Graph] Initializer onnx::LSTM_87 appears in graph inputs and will not be treated as constant value/weight. This may prevent some of the graph optimizations, like const folding. Move it out of graph inputs if there is no need to override it, by either re-generating the model with latest exporter/converter or with the tool onnxruntime/tools/python/remove_initializer_from_input.py.\n" + "2024-09-12 12:18:53.086326293 [W:onnxruntime:, graph.cc:1283 Graph] Initializer onnx::LSTM_87 appears in graph inputs and will not be treated as constant value/weight. This may prevent some of the graph optimizations, like const folding. Move it out of graph inputs if there is no need to override it, by either re-generating the model with latest exporter/converter or with the tool onnxruntime/tools/python/remove_initializer_from_input.py.\n" ] } ], @@ -1115,7 +1115,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:78: UserWarning: Keyword arguments are being passed but they not being used.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:55: UserWarning: Keyword arguments are being passed but they not being used.\n", " warn('Keyword arguments are being passed but they not being used.')\n" ] } @@ -1191,7 +1191,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:78: UserWarning: Keyword arguments are being passed but they not being used.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:55: UserWarning: Keyword arguments are being passed but they not being used.\n", " warn('Keyword arguments are being passed but they not being used.')\n" ] } @@ -1268,7 +1268,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:78: UserWarning: Keyword arguments are being passed but they not being used.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:55: UserWarning: Keyword arguments are being passed but they not being used.\n", " warn('Keyword arguments are being passed but they not being used.')\n" ] } @@ -1345,7 +1345,7 @@ "name": "stderr", "output_type": "stream", "text": [ - "/scratch/fabian/brevitas/src/brevitas/nn/mixin/base.py:78: UserWarning: Keyword arguments are being passed but they not being used.\n", + "/proj/xlabs/users/nfraser/opt/miniforge3/envs/20231115_brv_pt1.13.1/lib/python3.10/site-packages/brevitas/nn/mixin/base.py:55: UserWarning: Keyword arguments are being passed but they not being used.\n", " warn('Keyword arguments are being passed but they not being used.')\n" ] } @@ -1420,17 +1420,7 @@ "cell_type": "code", "execution_count": 32, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "[W shape_type_inference.cpp:1974] Warning: The shape inference of onnx.brevitas::QuantLSTMCell type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)\n", - "[W shape_type_inference.cpp:1974] Warning: The shape inference of onnx.brevitas::QuantLSTMCell type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)\n", - "[W shape_type_inference.cpp:1974] Warning: The shape inference of onnx.brevitas::QuantLSTMCell type is missing, so it may result in wrong shape inference for the exported graph. Please consider adding it in symbolic function. (function UpdateReliable)\n" - ] - } - ], + "outputs": [], "source": [ "import torch\n", "from brevitas.nn import QuantLSTM\n", @@ -1566,7 +1556,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.10.13" }, "vscode": { "interpreter": {