eftnas-s2-bert-medium-cola/compressed_graph.dot

strict digraph  {
"0 /nncf_model_input_0";
"1 /nncf_model_input_1";
"2 /nncf_model_input_2";
"3 BertForSequenceClassification/BertModel[bert]/__getitem___0";
"4 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[word_embeddings]/embedding_0";
"5 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[token_type_embeddings]/embedding_0";
"6 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__add___0";
"7 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[position_embeddings]/embedding_0";
"8 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__iadd___0";
"9 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"10 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/Dropout[dropout]/dropout_0";
"11 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"12 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"13 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"14 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"15 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"16 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"17 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"18 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"19 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"20 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"21 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"22 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"23 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"24 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"25 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"26 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"27 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"28 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"29 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"30 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"31 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"32 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"33 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"34 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"35 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"36 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"37 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"38 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"39 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"40 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"41 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"42 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"43 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"44 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"45 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"46 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"47 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"48 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"49 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/Dropout[dropout]/dropout_0";
"50 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/__add___0";
"51 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"52 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"53 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"54 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"55 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"56 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"57 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"58 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"59 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"60 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"61 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"62 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"63 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"64 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"65 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"66 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"67 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"68 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"69 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"70 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"71 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"72 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"73 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"74 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"75 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"76 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"77 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"78 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"79 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"80 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"81 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"82 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"83 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"84 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"85 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"86 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"87 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"88 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"89 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"90 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/Dropout[dropout]/dropout_0";
"91 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/__add___0";
"92 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"93 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"94 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"95 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"96 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"97 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"98 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"99 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"100 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"101 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"102 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"103 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"104 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"105 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"106 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"107 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"108 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"109 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"110 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"111 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"112 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"113 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"114 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"115 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"116 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"117 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"118 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"119 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"120 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"121 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"122 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"123 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"124 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"125 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"126 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"127 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"128 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"129 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"130 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"131 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/Dropout[dropout]/dropout_0";
"132 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/__add___0";
"133 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"134 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"135 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"136 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"137 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"138 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"139 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"140 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"141 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"142 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"143 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"144 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"145 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"146 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"147 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"148 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"149 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"150 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"151 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"152 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"153 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"154 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"155 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"156 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"157 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"158 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"159 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"160 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"161 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"162 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"163 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"164 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"165 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"166 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"167 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"168 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"169 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"170 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"171 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"172 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/Dropout[dropout]/dropout_0";
"173 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/__add___0";
"174 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"175 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"176 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"177 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"178 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"179 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"180 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"181 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"182 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"183 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"184 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"185 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"186 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"187 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"188 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"189 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"190 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"191 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"192 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"193 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"194 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"195 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"196 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"197 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"198 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"199 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"200 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"201 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"202 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"203 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"204 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"205 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"206 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"207 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"208 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"209 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"210 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"211 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"212 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"213 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/Dropout[dropout]/dropout_0";
"214 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/__add___0";
"215 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"216 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"217 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"218 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"219 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"220 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"221 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"222 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"223 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"224 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"225 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"226 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"227 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"228 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"229 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"230 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"231 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"232 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"233 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"234 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"235 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"236 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"237 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"238 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"239 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"240 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"241 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"242 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"243 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"244 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"245 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"246 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"247 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"248 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"249 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"250 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"251 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"252 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"253 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"254 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/Dropout[dropout]/dropout_0";
"255 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/__add___0";
"256 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"257 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"258 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"259 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"260 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"261 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"262 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"263 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"264 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"265 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"266 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"267 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"268 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"269 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"270 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"271 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"272 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"273 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"274 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"275 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"276 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"277 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"278 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"279 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"280 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"281 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"282 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"283 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"284 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"285 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"286 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"287 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"288 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"289 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"290 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"291 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"292 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"293 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"294 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"295 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/Dropout[dropout]/dropout_0";
"296 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/__add___0";
"297 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"298 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"299 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"300 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0";
"301 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"302 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"303 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0";
"304 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_0";
"305 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_0";
"306 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"307 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"308 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0";
"309 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_1";
"310 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_1";
"311 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_2";
"312 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_2";
"313 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/transpose_0";
"314 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_0";
"315 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0";
"316 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/__add___0";
"317 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/softmax_0";
"318 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0";
"319 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_1";
"320 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_3";
"321 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0";
"322 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_3";
"323 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"324 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"325 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0";
"326 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0";
"327 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/__add___0";
"328 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"329 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"330 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"331 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0";
"332 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0";
"333 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0";
"334 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1";
"335 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/linear_0";
"336 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/Dropout[dropout]/dropout_0";
"337 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/__add___0";
"338 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0";
"339 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/__getitem___0";
"340 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/NNCFLinear[dense]/linear_0";
"341 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/Tanh[activation]/tanh_0";
"342 BertForSequenceClassification/Dropout[dropout]/dropout_0";
"343 BertForSequenceClassification/NNCFLinear[classifier]/linear_0";
"344 /nncf_model_output_0";
"0 /nncf_model_input_0" -> "4 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[word_embeddings]/embedding_0"  [label="(1, 128) \n0 -> 0", style=dashed];
"1 /nncf_model_input_1" -> "3 BertForSequenceClassification/BertModel[bert]/__getitem___0"  [label="(1, 128) \n0 -> 0", style=dashed];
"2 /nncf_model_input_2" -> "5 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[token_type_embeddings]/embedding_0"  [label="(1, 128) \n0 -> 0", style=dashed];
"4 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[word_embeddings]/embedding_0" -> "6 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"5 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[token_type_embeddings]/embedding_0" -> "6 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"6 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__add___0" -> "8 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__iadd___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"7 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFEmbedding[position_embeddings]/embedding_0" -> "8 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__iadd___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"8 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/__iadd___0" -> "9 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"9 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "10 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"10 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/Dropout[dropout]/dropout_0" -> "13 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"10 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/Dropout[dropout]/dropout_0" -> "16 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"10 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/Dropout[dropout]/dropout_0" -> "21 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"10 BertForSequenceClassification/BertModel[bert]/BertEmbeddings[embeddings]/Dropout[dropout]/dropout_0" -> "40 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"11 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "13 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"12 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "13 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"13 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "24 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"14 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "16 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"15 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "16 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"16 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "17 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"17 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "18 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"18 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "26 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"19 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "21 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"20 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "21 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"21 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "22 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"22 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "23 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"23 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "32 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"24 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "25 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"25 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "27 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"26 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "27 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"27 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "28 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"28 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "29 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"29 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "30 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"30 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "31 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"31 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "32 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"32 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "33 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"33 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "34 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"34 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "35 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"35 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "38 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"36 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "38 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"37 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "38 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"38 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "39 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"39 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "40 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"40 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "41 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"41 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "44 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"41 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "50 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"42 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "44 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"43 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "44 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"44 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "45 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"45 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "48 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"46 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "48 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"47 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "48 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"48 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "49 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"49 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "50 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"50 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/__add___0" -> "51 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"51 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "54 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"51 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "57 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"51 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "62 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"51 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[0]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "81 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"52 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "54 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"53 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "54 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"54 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "65 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"55 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "57 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"56 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "57 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"57 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "58 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"58 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "59 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"59 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "67 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"60 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "62 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"61 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "62 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"62 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "63 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"63 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "64 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"64 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "73 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"65 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "66 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"66 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "68 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"67 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "68 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"68 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "69 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"69 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "70 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"70 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "71 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"71 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "72 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"72 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "73 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"73 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "74 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"74 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "75 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"75 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "76 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"76 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "79 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"77 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "79 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"78 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "79 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"79 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "80 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"80 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "81 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"81 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "82 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"82 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "85 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"82 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "91 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"83 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "85 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"84 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "85 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"85 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "86 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"86 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "89 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"87 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "89 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"88 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "89 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"89 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "90 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"90 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "91 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"91 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/__add___0" -> "92 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"92 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "95 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"92 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "98 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"92 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "103 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"92 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[1]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "122 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"93 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "95 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"94 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "95 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"95 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "106 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"96 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "98 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"97 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "98 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"98 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "99 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"99 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "100 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"100 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "108 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"101 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "103 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"102 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "103 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"103 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "104 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"104 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "105 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"105 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "114 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"106 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "107 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"107 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "109 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"108 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "109 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"109 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "110 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"110 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "111 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"111 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "112 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"112 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "113 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"113 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "114 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"114 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "115 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"115 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "116 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"116 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "117 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"117 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "120 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"118 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "120 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"119 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "120 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"120 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "121 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"121 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "122 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"122 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "123 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"123 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "126 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"123 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "132 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"124 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "126 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"125 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "126 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"126 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "127 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"127 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "130 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"128 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "130 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"129 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "130 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"130 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "131 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"131 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "132 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"132 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/__add___0" -> "133 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"133 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "136 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"133 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "139 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"133 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "144 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"133 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[2]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "163 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"134 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "136 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"135 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "136 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"136 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "147 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"137 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "139 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"138 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "139 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"139 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "140 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"140 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "141 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"141 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "149 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"142 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "144 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"143 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "144 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"144 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "145 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"145 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "146 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"146 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "155 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"147 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "148 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"148 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "150 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"149 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "150 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"150 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "151 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"151 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "152 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"152 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "153 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"153 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "154 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"154 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "155 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"155 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "156 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"156 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "157 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"157 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "158 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"158 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "161 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"159 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "161 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"160 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "161 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"161 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "162 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"162 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "163 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"163 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "164 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"164 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "167 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"164 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "173 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"165 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "167 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"166 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "167 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"167 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "168 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"168 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "171 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"169 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "171 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"170 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "171 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"171 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "172 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"172 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "173 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"173 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/__add___0" -> "174 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"174 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "177 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"174 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "180 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"174 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "185 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"174 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[3]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "204 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"175 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "177 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"176 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "177 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"177 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "188 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"178 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "180 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"179 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "180 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"180 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "181 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"181 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "182 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"182 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "190 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"183 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "185 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"184 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "185 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"185 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "186 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"186 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "187 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"187 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "196 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"188 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "189 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"189 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "191 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"190 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "191 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"191 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "192 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"192 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "193 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"193 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "194 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"194 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "195 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"195 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "196 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"196 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "197 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"197 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "198 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"198 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "199 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"199 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "202 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"200 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "202 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"201 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "202 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"202 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "203 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"203 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "204 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"204 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "205 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"205 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "208 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"205 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "214 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"206 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "208 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"207 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "208 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"208 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "209 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"209 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "212 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"210 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "212 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"211 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "212 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"212 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "213 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"213 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "214 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"214 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/__add___0" -> "215 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"215 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "218 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"215 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "221 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"215 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "226 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"215 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[4]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "245 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"216 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "218 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"217 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "218 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"218 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "229 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"219 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "221 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"220 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "221 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"221 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "222 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"222 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "223 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"223 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "231 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"224 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "226 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"225 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "226 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"226 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "227 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"227 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "228 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"228 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "237 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"229 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "230 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"230 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "232 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"231 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "232 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"232 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "233 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"233 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "234 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"234 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "235 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"235 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "236 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"236 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "237 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"237 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "238 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"238 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "239 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"239 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "240 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"240 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "243 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"241 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "243 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"242 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "243 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"243 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "244 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"244 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "245 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"245 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "246 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"246 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "249 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"246 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "255 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"247 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "249 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"248 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "249 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"249 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "250 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"250 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "253 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"251 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "253 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"252 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "253 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"253 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "254 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"254 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "255 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"255 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/__add___0" -> "256 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"256 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "259 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"256 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "262 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"256 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "267 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"256 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[5]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "286 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"257 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "259 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"258 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "259 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"259 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "270 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"260 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "262 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"261 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "262 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"262 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "263 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"263 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "264 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"264 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "272 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"265 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "267 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"266 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "267 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"267 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "268 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"268 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "269 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"269 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "278 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"270 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "271 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"271 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "273 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"272 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "273 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"273 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "274 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"274 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "275 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"275 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "276 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"276 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "277 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"277 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "278 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"278 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "279 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"279 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "280 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"280 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "281 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"281 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "284 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"282 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "284 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"283 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "284 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"284 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "285 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"285 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "286 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"286 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "287 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"287 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "290 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"287 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "296 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"288 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "290 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"289 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "290 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"290 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "291 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"291 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "294 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"292 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "294 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"293 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "294 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"294 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "295 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"295 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "296 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"296 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/__add___0" -> "297 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"297 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "300 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"297 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "303 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"297 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "308 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"297 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[6]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "327 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"298 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "300 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"299 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "300 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"300 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[query]/linear_0" -> "311 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_2"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"301 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "303 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"302 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "303 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"303 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[key]/linear_0" -> "304 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"304 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_0" -> "305 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"305 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_0" -> "313 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/transpose_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"306 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "308 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"307 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "308 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"308 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/NNCFLinear[value]/linear_0" -> "309 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_1"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"309 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_1" -> "310 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_1"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"310 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_1" -> "319 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 64) \n0 -> 1", style=solid];
"311 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_2" -> "312 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_2"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"312 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_2" -> "314 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"313 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/transpose_0" -> "314 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_0"  [label="(1, 8, 64, 128) \n0 -> 1", style=solid];
"314 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_0" -> "315 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"315 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/__truediv___0" -> "316 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/__add___0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"316 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/__add___0" -> "317 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/softmax_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"317 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/softmax_0" -> "318 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"318 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/Dropout[dropout]/dropout_0" -> "319 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_1"  [label="(1, 8, 128, 128) \n0 -> 0", style=solid];
"319 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/matmul_1" -> "320 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_3"  [label="(1, 8, 128, 64) \n0 -> 0", style=solid];
"320 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/permute_3" -> "321 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"321 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/contiguous_0" -> "322 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_3"  [label="(1, 128, 8, 64) \n0 -> 0", style=solid];
"322 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfAttention[self]/view_3" -> "325 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"323 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "325 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 512) \n0 -> 1", style=solid];
"324 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "325 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"325 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLinear[dense]/linear_0" -> "326 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"326 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/Dropout[dropout]/dropout_0" -> "327 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"327 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/__add___0" -> "328 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"328 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "331 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"328 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertAttention[attention]/BertSelfOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "337 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 1", style=solid];
"329 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "331 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048, 512) \n0 -> 1", style=solid];
"330 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "331 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0"  [label="(2048,) \n0 -> 2", style=solid];
"331 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/NNCFLinear[dense]/linear_0" -> "332 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"332 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertIntermediate[intermediate]/GELUActivation[intermediate_act_fn]/gelu_0" -> "335 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(1, 128, 2048) \n0 -> 0", style=solid];
"333 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_0" -> "335 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512, 2048) \n0 -> 1", style=solid];
"334 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/ModuleDict[pre_ops]/UpdateWeightAndBias[0]/MovementSparsifier[op]/apply_binary_mask_1" -> "335 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/linear_0"  [label="(512,) \n0 -> 2", style=solid];
"335 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLinear[dense]/linear_0" -> "336 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/Dropout[dropout]/dropout_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"336 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/Dropout[dropout]/dropout_0" -> "337 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/__add___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"337 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/__add___0" -> "338 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"338 BertForSequenceClassification/BertModel[bert]/BertEncoder[encoder]/ModuleList[layer]/BertLayer[7]/BertOutput[output]/NNCFLayerNorm[LayerNorm]/layer_norm_0" -> "339 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/__getitem___0"  [label="(1, 128, 512) \n0 -> 0", style=solid];
"339 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/__getitem___0" -> "340 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/NNCFLinear[dense]/linear_0"  [label="(1, 512) \n0 -> 0", style=solid];
"340 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/NNCFLinear[dense]/linear_0" -> "341 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/Tanh[activation]/tanh_0"  [label="(1, 512) \n0 -> 0", style=solid];
"341 BertForSequenceClassification/BertModel[bert]/BertPooler[pooler]/Tanh[activation]/tanh_0" -> "342 BertForSequenceClassification/Dropout[dropout]/dropout_0"  [label="(1, 512) \n0 -> 0", style=solid];
"342 BertForSequenceClassification/Dropout[dropout]/dropout_0" -> "343 BertForSequenceClassification/NNCFLinear[classifier]/linear_0"  [label="(1, 512) \n0 -> 0", style=solid];
"343 BertForSequenceClassification/NNCFLinear[classifier]/linear_0" -> "344 /nncf_model_output_0"  [label="(1, 2) \n0 -> 0", style=solid];
}