diff --git a/examples/movement-pruning/Saving_PruneBERT.ipynb b/examples/movement-pruning/Saving_PruneBERT.ipynb index f0c68433f84..01fcd3257cd 100644 --- a/examples/movement-pruning/Saving_PruneBERT.ipynb +++ b/examples/movement-pruning/Saving_PruneBERT.ipynb @@ -63,532 +63,7 @@ "cell_type": "code", "execution_count": 2, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "BertForQuestionAnswering(\n", - " (bert): BertModel(\n", - " (embeddings): BertEmbeddings(\n", - " (word_embeddings): Embedding(30522, 768, padding_idx=0)\n", - " (position_embeddings): Embedding(512, 768)\n", - " (token_type_embeddings): Embedding(2, 768)\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (encoder): BertEncoder(\n", - " (layer): ModuleList(\n", - " (0): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (1): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (2): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (3): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (4): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (5): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (6): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (7): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (8): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (9): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (10): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (11): BertLayer(\n", - " (attention): BertAttention(\n", - " (self): BertSelfAttention(\n", - " (query): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (key): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (value): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " (output): BertSelfOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " (intermediate): BertIntermediate(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=3072\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " )\n", - " (output): BertOutput(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=3072, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)\n", - " (dropout): Dropout(p=0.1, inplace=False)\n", - " )\n", - " )\n", - " )\n", - " )\n", - " (pooler): BertPooler(\n", - " (dense): DynamicQuantizedLinear(\n", - " in_features=768, out_features=768\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - " (activation): Tanh()\n", - " )\n", - " )\n", - " (qa_outputs): DynamicQuantizedLinear(\n", - " in_features=768, out_features=2\n", - " (_packed_params): LinearPackedParams()\n", - " )\n", - ")\n" - ] - } - ], + "outputs": [], "source": [ "# Load fine-pruned model and quantize the model\n", "\n", @@ -605,7 +80,7 @@ " },\n", " dtype=torch.qint8,\n", " )\n", - "print(quantized_model)\n", + "# print(quantized_model)\n", "\n", "qtz_st = quantized_model.state_dict()" ] @@ -910,745 +385,7 @@ "cell_type": "code", "execution_count": 8, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.0.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.0.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.0.attention.self.key.scale\n", - "Unpack bert.encoder.layer.0.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.0.attention.self.query.scale\n", - "Unpack bert.encoder.layer.0.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.0.attention.self.value.scale\n", - "Unpack bert.encoder.layer.0.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.0.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.0.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.0.output.dense.scale\n", - "Unpack bert.encoder.layer.0.output.dense.zero_point\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.1.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.1.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.1.attention.self.key.scale\n", - "Unpack bert.encoder.layer.1.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.1.attention.self.query.scale\n", - "Unpack bert.encoder.layer.1.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.1.attention.self.value.scale\n", - "Unpack bert.encoder.layer.1.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.1.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.1.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.1.output.dense.scale\n", - "Unpack bert.encoder.layer.1.output.dense.zero_point\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.10.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.10.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.10.attention.self.key.scale\n", - "Unpack bert.encoder.layer.10.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.10.attention.self.query.scale\n", - "Unpack bert.encoder.layer.10.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.10.attention.self.value.scale\n", - "Unpack bert.encoder.layer.10.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.10.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.10.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.10.output.dense.scale\n", - "Unpack bert.encoder.layer.10.output.dense.zero_point\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.11.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.11.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.11.attention.self.key.scale\n", - "Unpack bert.encoder.layer.11.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.11.attention.self.query.scale\n", - "Unpack bert.encoder.layer.11.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.11.attention.self.value.scale\n", - "Unpack bert.encoder.layer.11.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.11.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.11.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.11.output.dense.scale\n", - "Unpack bert.encoder.layer.11.output.dense.zero_point\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.2.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.2.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.2.attention.self.key.scale\n", - "Unpack bert.encoder.layer.2.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.2.attention.self.query.scale\n", - "Unpack bert.encoder.layer.2.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.2.attention.self.value.scale\n", - "Unpack bert.encoder.layer.2.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.2.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.2.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.2.output.dense.scale\n", - "Unpack bert.encoder.layer.2.output.dense.zero_point\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.3.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.3.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.3.attention.self.key.scale\n", - "Unpack bert.encoder.layer.3.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.3.attention.self.query.scale\n", - "Unpack bert.encoder.layer.3.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.3.attention.self.value.scale\n", - "Unpack bert.encoder.layer.3.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.3.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.3.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.3.output.dense.scale\n", - "Unpack bert.encoder.layer.3.output.dense.zero_point\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.4.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.4.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.4.attention.self.key.scale\n", - "Unpack bert.encoder.layer.4.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.4.attention.self.query.scale\n", - "Unpack bert.encoder.layer.4.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.4.attention.self.value.scale\n", - "Unpack bert.encoder.layer.4.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.4.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.4.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.4.output.dense.scale\n", - "Unpack bert.encoder.layer.4.output.dense.zero_point\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.5.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.5.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.5.attention.self.key.scale\n", - "Unpack bert.encoder.layer.5.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.5.attention.self.query.scale\n", - "Unpack bert.encoder.layer.5.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.5.attention.self.value.scale\n", - "Unpack bert.encoder.layer.5.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.5.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.5.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.5.output.dense.scale\n", - "Unpack bert.encoder.layer.5.output.dense.zero_point\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.6.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.6.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.6.attention.self.key.scale\n", - "Unpack bert.encoder.layer.6.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.6.attention.self.query.scale\n", - "Unpack bert.encoder.layer.6.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.6.attention.self.value.scale\n", - "Unpack bert.encoder.layer.6.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.6.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.6.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.6.output.dense.scale\n", - "Unpack bert.encoder.layer.6.output.dense.zero_point\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.7.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.7.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.7.attention.self.key.scale\n", - "Unpack bert.encoder.layer.7.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.7.attention.self.query.scale\n", - "Unpack bert.encoder.layer.7.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.7.attention.self.value.scale\n", - "Unpack bert.encoder.layer.7.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.7.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.7.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.7.output.dense.scale\n", - "Unpack bert.encoder.layer.7.output.dense.zero_point\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.8.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.8.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.8.attention.self.key.scale\n", - "Unpack bert.encoder.layer.8.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.8.attention.self.query.scale\n", - "Unpack bert.encoder.layer.8.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.8.attention.self.value.scale\n", - "Unpack bert.encoder.layer.8.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.8.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.8.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.8.output.dense.scale\n", - "Unpack bert.encoder.layer.8.output.dense.zero_point\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.9.attention.output.dense.scale\n", - "Unpack bert.encoder.layer.9.attention.output.dense.zero_point\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.9.attention.self.key.scale\n", - "Unpack bert.encoder.layer.9.attention.self.key.zero_point\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.9.attention.self.query.scale\n", - "Unpack bert.encoder.layer.9.attention.self.query.zero_point\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.9.attention.self.value.scale\n", - "Unpack bert.encoder.layer.9.attention.self.value.zero_point\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.9.intermediate.dense.scale\n", - "Unpack bert.encoder.layer.9.intermediate.dense.zero_point\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.weight.scale\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.weight.zero_point\n", - "Unpack bert.encoder.layer.9.output.dense.scale\n", - "Unpack bert.encoder.layer.9.output.dense.zero_point\n", - "Unpack bert.pooler.dense._packed_params.weight.int_repr.shape\n", - "Unpack bert.pooler.dense._packed_params.weight.scale\n", - "Unpack bert.pooler.dense._packed_params.weight.zero_point\n", - "Unpack bert.pooler.dense.scale\n", - "Unpack bert.pooler.dense.zero_point\n", - "Unpack qa_outputs._packed_params.weight.int_repr.shape\n", - "Unpack qa_outputs._packed_params.weight.scale\n", - "Unpack qa_outputs._packed_params.weight.zero_point\n", - "Unpack qa_outputs.scale\n", - "Unpack qa_outputs.zero_point\n", - "Unpack bert.embeddings.LayerNorm.bias\n", - "Unpack bert.embeddings.LayerNorm.weight\n", - "Unpack bert.embeddings.position_embeddings.weight\n", - "Unpack bert.embeddings.token_type_embeddings.weight\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Unpack bert.embeddings.word_embeddings.weight\n", - "Unpack bert.encoder.layer.0.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.0.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.0.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.0.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.0.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.0.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.0.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.0.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.0.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.0.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.1.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.1.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.1.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.1.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.1.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.1.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.1.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.1.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.1.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.1.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.10.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.10.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.10.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.10.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.10.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.10.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.10.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.10.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.10.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.10.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.11.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.11.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.11.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.11.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.11.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.11.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.11.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.11.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.11.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.11.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.2.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.2.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.2.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.2.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.2.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.2.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.2.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.2.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.2.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.2.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.3.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.3.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.3.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.3.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.3.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.3.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.3.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.3.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.3.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.3.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.4.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.4.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.4.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.4.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.4.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.4.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.4.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.4.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.4.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.4.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.5.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.5.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.5.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.5.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.5.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.bias\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.5.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.5.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.5.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.5.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.5.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.6.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.6.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.6.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.6.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.6.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.6.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.6.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.6.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.6.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.6.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.7.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.7.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.7.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.7.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.7.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.7.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.7.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.7.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.7.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.7.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.8.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.8.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.8.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.8.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.8.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.8.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.8.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.8.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.8.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.8.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.9.attention.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.9.attention.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.9.attention.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.bias\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.9.attention.self.key._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.bias\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.9.attention.self.query._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.bias\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.9.attention.self.value._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.9.intermediate.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.encoder.layer.9.output.LayerNorm.bias\n", - "Unpack bert.encoder.layer.9.output.LayerNorm.weight\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.bias\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.encoder.layer.9.output.dense._packed_params.weight.int_repr.indptr\n", - "Unpack bert.pooler.dense._packed_params.bias\n", - "Unpack bert.pooler.dense._packed_params.weight.int_repr.data\n", - "Unpack bert.pooler.dense._packed_params.weight.int_repr.indices\n", - "Unpack bert.pooler.dense._packed_params.weight.int_repr.indptr\n", - "Unpack qa_outputs._packed_params.bias\n", - "Unpack qa_outputs._packed_params.weight.int_repr.data\n", - "Unpack qa_outputs._packed_params.weight.int_repr.indices\n", - "Unpack qa_outputs._packed_params.weight.int_repr.indptr\n" - ] - } - ], + "outputs": [], "source": [ "# Reconstruct the elementary state dict\n", "\n", @@ -1670,7 +407,7 @@ " else:\n", " attr_param = torch.tensor(attr_param)\n", " reconstructed_elementary_qtz_st[attr_name] = attr_param\n", - " print(f\"Unpack {attr_name}\")\n", + " # print(f\"Unpack {attr_name}\")\n", " \n", "# Get the tensors/arrays\n", "for data_name, data_param in hf.items():\n", @@ -1683,7 +420,7 @@ " if \"indices\" in data_name:\n", " data_param = np.array(data_param, dtype=np.int32)\n", " reconstructed_elementary_qtz_st[data_name] = data_param\n", - " print(f\"Unpack {data_name}\")\n", + " # print(f\"Unpack {data_name}\")\n", " \n", "\n", "hf.close()" @@ -1818,7 +555,15 @@ "cell_type": "code", "execution_count": 13, "metadata": {}, - "outputs": [], + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Sanity check passed\n" + ] + } + ], "source": [ "# Sanity checks on the infernce\n", "\n", @@ -1831,7 +576,8 @@ " y_reconstructed = reconstructed_qtz_model(input_ids=inputs, attention_mask=mask)[0]\n", " y = quantized_model(input_ids=inputs, attention_mask=mask)[0]\n", " \n", - " assert torch.all(torch.eq(y, y_reconstructed))" + " assert torch.all(torch.eq(y, y_reconstructed))\n", + "print(\"Sanity check passed\")" ] }, {