Project dependencies may have API risk issues #242
Description
Hi, In mt-dnn, inappropriate dependency versioning constraints can cause risks.
Below are the dependencies and version constraints that the project is using
numpy
torch==1.5.0
tqdm
colorlog
boto3
pytorch-pretrained-bert==v0.6.0
regex
scikit-learn
pyyaml
pytest
sentencepiece
tensorboardX
tensorboard
future
apex
seqeval==0.0.12
transformers==4.6.0
The version constraint == will introduce the risk of dependency conflicts because the scope of dependencies is too strict.
The version constraint No Upper Bound and * will introduce the risk of the missing API Error because the latest version of the dependencies may remove some APIs.
After further analysis, in this project,
The version constraint of dependency numpy can be changed to >=1.8.0,<=1.23.0rc3.
The version constraint of dependency tqdm can be changed to >=4.36.0,<=4.64.0.
The version constraint of dependency pytorch-pretrained-bert can be changed to >=0.1.1,<=0.3.0.
The version constraint of dependency scikit-learn can be changed to >=0.15.0,<=0.20.4.
The version constraint of dependency future can be changed to >=0.12.0,<=0.18.2.
The version constraint of dependency transformers can be changed to >=2.0.0,<=4.1.1.
The above modification suggestions can reduce the dependency conflicts as much as possible,
and introduce the latest version as much as possible without calling Error in the projects.
The invocation of the current project includes all the following methods.
The calling methods from the numpy
numpy.ma.masked_array
The calling methods from the tqdm
tqdm.tqdm
The calling methods from the pytorch-pretrained-bert
pytorch_pretrained_bert.modeling.BertLayerNorm
The calling methods from the scikit-learn
sklearn.metrics.f1_score random.choice sklearn.metrics.accuracy_score sklearn.metrics.matthews_corrcoef sklearn.metrics.roc_auc_score sklearn.metrics.confusion_matrix
The calling methods from the future
time.strftime datetime.datetime.now
The calling methods from the transformers
transformers.AutoTokenizer.from_pretrained
The calling methods from the all methods
plt.clf _make_qid_to_has_ans grad.abs.add_ sorted.append torch.nn.Tanh score.reshape.tolist.numpy os.path.isdir torch.erf logging.StreamHandler.setFormatter self._task_def_dic.items layer_module module.register_forward_pre_hook subprocess.call.strip iter make_qid_to_has_ans.items self._prepare_model_input rnn_cls m.contiguous src.items set data_utils.log_wrapper.create_logger.warning numpy.ma.masked_array extra_indices.tolist.tolist mask.size output_dir.os.path.join.open.read self.shadow.items find_best_thresh self.qkv_head_dim.self.num_heads.bsz.src_len.k.contiguous.view.transpose find_all_best_thresh min truncate_seq_pair data.get.str.lower score.numpy.numpy set_config self.dropout.transpose transformers.get_polynomial_decay_schedule_with_warmup key_padding_mask.unsqueeze.unsqueeze input_lengths.append self.tok2ind.keys final_predict.append passage.replace.replace prefix.format.opt.get.upper logging.StreamHandler f.readlines logging.StreamHandler.setLevel _get_raw_scores torch.nn.functional.cross_entropy line.strip.split tqdm.tqdm prepare_validation_features AttentionWrapper range format.startswith self._tokenizer.vocab.keys tokens_b.extend scores.data.masked_fill_ qb.new_dict.cpu.numpy rnn_type.upper.upper delattr score.numpy.tolist data_utils.task_def.EncoderModelType mlm_p.view.size shutil.rmtree load_stsb self.AdamaxW.super.__init__ self.kd_task_loss_criterion.append self.dropout.bmm partial_feature data_utils.log_wrapper.create_logger.error x.size self.SmartPerturbation.super.__init__ torch.tanh self.eff_perturb.update copy.deepcopy.item rows.append key_padding_mask.unsqueeze score.reshape.tolist self.linear.contiguous test_prefix.lower self.score_func load_qqp self.SimpleFlatSim.super.__init__ self.BilinearSum.super.__init__ make_qid_to_has_ans.values json.dumps extract_answers_from_features x_flat.self.linear.view self.rnn.new os.path.exists TaskDef text.lower round self.Pooler.super.__init__ self.decoder self.dropout copy.deepcopy.extend self.DropoutWrapper.super.__init__ self.y_linear logging.FileHandler inputs.append data_utils.vocab.Vocabulary SanPooler numpy.random.seed x1.contiguous.view args.train_datasets.split self._global_map.get i.multi_task_train_data.len.i.start.datetime.now.str.split search_index torch.float32.target.detach.F.softmax.sqrt_ kb.new_dict.cpu.numpy mt_dnn.batcher.Collater torch.device open key.transpose.size masked_lm_positions.append collections.OrderedDict torch.max torch.sum text.split random.randint self.dropout.contiguous torch.unbind x1_flat.self.proj_1.self.f.view.bmm self.qkv_head_dim.self.num_heads.bsz.src_len.v.contiguous.view.transpose.contiguous experiments.superglue.superglue_utils.save all_encoder_layers.append y.unsqueeze.expand_as.unsqueeze self.optimizer.zero_grad data_utils.tokenizer_utils.create_tokenizer self.encode qa.get mt_dnn.batcher.DistTaskDataset batch_meta.batch_data.detach.cpu.numpy.tolist self.classifier sklearn.metrics.roc_auc_score self.SANBertNetwork.super.__init__ grad.abs.max self.optimizer.load_state_dict self.qkv_head_dim.self.num_heads.bsz.tgt_len.q.contiguous.view.transpose self.init self.setup self.qkv_head_dim.self.num_heads.bsz.src_len.k.contiguous.view.transpose.transpose self.alpha.data.new torch.autograd.grad common.activation kb.new_dict.cpu load_json super labels.append tasks.get_task_obj.train_forward super.__init__ beta2.exp_inf.mul_.unsqueeze logits.size.logits.data.new.zero_.torch.diag.byte train_datasets.append torch.nn.functional.kl_div self.f self.train_loss.update seqeval.metrics.classification_report FlatSimV2 embed.detach.abs.mean subprocess.check_output scipy.stats.spearmanr x.torch.std.expand_as numpy.argmax make_eval_dict mt_dnn.batcher.DistMultiTaskBatchSampler score.reshape.tolist.reshape mt_dnn.model.MTDNNModel.update k.contiguous.view token.startswith p.transpose module.my_optim.weight_norm noise.detach.requires_grad_ self.DotProduct.super.__init__ mt_dnn.batcher.Collater.patch_data load_mrpc SimpleFlatSim eff_noise.detach plt.hist adv_logits.view.view load_copa_mtdnn run_precision_recall_analysis trim torch.zeros_like batch_info.pin_memory.to self.DeepAttentionWrapper.super.__init__ target.contiguous.view.view int self.parameters self.qkv_head_dim.self.num_heads.bsz.src_len.k.contiguous.view.transpose.contiguous torch.manual_seed merge_eval x2_flat.self.proj_2.self.f.view.transpose scores.items shutil.copytree plot_pr_curve self.__str__ grad.sign json.loads.get time.strftime BilinearFlatSim compute_f1 reduce_features_to_examples.append closure end.numpy.tolist.contiguous data_utils.load_data self.bert create_bins self.SANClassifier.super.__init__ get_tokens logging.FileHandler.setLevel self._tokenizer.convert_tokens_to_ids.get y.x.torch.abs.y.x.y.x.torch.cat.contiguous x.size.idx.mask.view.expand_as self.rnn.size embedding_weights.size torch.nn.functional.mse_loss uid.split torch.LongTensor sample.get.strip os.getcwd self._rnn att_scores.F.softmax.unsqueeze temp_answers.append test_prepro_std self.MLPSelfAttn.super.__init__ self.FlatSimV2.super.__init__ torch.nn.ModuleList main load_data os.mkdir self.compute_weight self._setup_lossmap setattr x1.size.x1.size.x1.contiguous.view.self.x_dot_linear.view.expand_as.bmm plt.savefig epsilon.epsilon.y.detach module.similarity.SelfAttnWrapper torch.distributed.all_reduce ground_truth.normalize_answer.split batch_meta.batch_data.size self.mnetwork.bert self.query_wsum prediction.normalize_answer.split dim.p.transpose._norm.transpose self._norm_grad torch.nn.Linear start_scores.squeeze.squeeze adv_lc model_config remove_punc self.config.update torch.isnan batch_meta.batch_data.detach para.strip tensorboardX.SummaryWriter.close data_utils.roberta_utils.update_roberta_keys sub_part.pin_memory.to start.data.cpu self.x_dot_linear NotImplementedError self.attn set.add sequence_outputs.append slen.idx.yidx.embeddings.tolist self.load model.to.to exp_avg.mul_ self.task_types.append self.encoder numpy.array task_id.self.scoring_list.split self.qkv_head_dim.self.num_heads.bsz.tgt_len.q.contiguous.view.transpose.contiguous task_id.self.scoring_list y.x.torch.abs.y.x.y.x.torch.cat.contiguous.view y.view.view module.common.activation self.score_func.size experiments.common_utils.dump_rows self.emb_val.update value.transpose.transpose list.extend tasks.get_task_obj.train_build_task_layer start.append vars.update transformers.get_cosine_schedule_with_warmup train_data_list.append transformers.get_linear_schedule_with_warmup predict.reshape.tolist dataset.get_task_id self._tokenizer.convert_tokens_to_ids end.append self.WeightNorm.super.__init__ self.linear idx.lines.strip torch.mean argparse.ArgumentParser.print_help mt_dnn.optim.AdamaxW score.np.argmax.reshape qb.new_dict.cpu self._dataset.get_task_id plt.xlim line.strip self.network.eval dump_data train_config.parse_args vw.new_dict.cpu end.numpy.tolist.numpy load alpha.unsqueeze.bmm collections.defaultdict self.num_hid.size.size.tmp_output.view.max dropout_layer question.strip self._get_param_groups bool newbatch.append s.normalize_answer.split out_f.write self.qkv_head_dim.self.num_heads.bsz.src_len.v.contiguous.view.transpose input_ids.index normalize_answer any key_padding_mask.size torch.nn.utils.weight_norm.unsqueeze machines.append tasks.get_task_obj.input_parse_label stable_kl pred.pop adv_loss.item tensor.to delta_grad.norm torch.nn.utils.weight_norm issubclass self.FlatSimilarityWrapper.super.__init__ e.to start.numpy.tolist.numpy score.data.cpu max x.att_scores.F.softmax.unsqueeze.torch.bmm.squeeze tasks.get_task_obj torch.nn.parallel.DistributedDataParallel self.config.get.backward self.Bilinear.super.__init__ task_id.self.task_loss_criterion logging.FileHandler.setFormatter float experiments.exp_def.TaskDefs.get_task_names load_multirc_mtdnn len.get result.strip.split model.to.parameters _norm self.dropout.size torch.norm hasattr torch.diag temp_1.pop self.LinearSelfAttn.super.__init__ data_utils.mrc_eval.squadv2_evaluate_func TypeError self.scalar.expand_as initialize_distributed task_cls LinearSelfAttn sorted start.numpy.tolist.contiguous torch.utils.data.DataLoader model_class.from_pretrained state.items enumerate sys.exit self.bert.generate self._setup_kd_lossmap self._gen_task_indices plt.title data_utils.mrc_eval.squadv1_evaluate_func weight.cpu join os.path.abspath numpy.max tasks.get_task_obj.test_predict target.size.target.view.float pdb.set_trace key.transpose.transpose torch.cat random.uniform literal_model_type.lower.lower exp_inf.new get_raw_scores self.adv_task_loss_criterion.append epsilon.epsilon.p.detach.log emb_val.item features.append x.torch.mean.expand_as new_data.zero_.size qw.new_dict.cpu load_mlm_data tok_len.batch_size.torch.BoolTensor.fill_ self.proj_1 torch.abs json.loads.split dump q.contiguous.view eff_noise.detach.abs metric_fn self.network.cuda batch_meta.get search_bin bais.cpu.numpy type_ids_list.append make_precision_recall_eval create_masked_lm_predictions attn_weights.data.masked_fill_ logging.Formatter y.x.torch.cat.contiguous.view self.proj ValueError logits.size.logits.data.new.zero_.torch.diag.byte.unsqueeze.expand_as i.tgt.extend query.replace time.gmtime load_mnli histogram_na_prob collections.Counter logit.size.logit.view.float.size generate_mask print_message local_task_idx.self._datasets.get_task_id torch.load.size part.pin_memory positions.append self.activation torch.pow assert_file_equal tasks.get_task_obj.train_prepare_label attn_weights.size.attn_weights.data.new.zero_.torch.diag.byte torch.isinf x1.size.x1.contiguous.view.self.x_dot_linear.view numpy.array_equal torch.nn.utils.clip_grad_norm_ grounds.items Vocabulary attn_weights.float.masked_fill self.embed_encode type cls mt_dnn.model.MTDNNModel.load mt_dnn.model.MTDNNModel.save self.__if_pair__ task.split format module._parameters.keys merge logits.detach logit.F.log_softmax.exp model.detach subprocess.call dropout_wrapper.DropoutWrapper eval config_class.from_dict isinstance scaled_loss.backward datetime.datetime.now self.dropout.view self._get_max_len kd_lc args.test_datasets.split dataset.items input_ids_list.append logits.size.logits.data.new.zero_.torch.diag.byte.unsqueeze tokenizer.vocab.keys load_record_mtdnn evaluate logging.getLogger.addHandler beta1.exp_avg.mul_.add_ load_sst batch_info.pin_memory locals qw.new_dict.cpu.numpy new_batch.append state.keys batch_data.append expected_file.open.read end_position.append laod_function FlatSim attn_weights.data.new torch.nn.functional.log_softmax i.self.attn_list self.embeddings process_data plt.xlabel load_xnli mt_dnn.model.MTDNNModel.cuda mt_dnn.batcher.MultiTaskBatchSampler scores.extend map score.contiguous.view self.rnn self.proj_2 vars data_utils.load_score_file.keys temp_1.append self._get_index_batches mt_dnn.model.MTDNNModel y.x.y.x.torch.cat.contiguous.view data_utils.metrics.calc_metrics MultiheadAttentionWrapper feature.example_id_to_index.features_per_example.append para.clone self.beta.expand_as opt.EncoderModelType.name.lower scores_for_ground_truths.append sentence.append self.pooler self.network.train numpy.zeros label.append scores_list.append os.getenv copy.deepcopy tokens.append p.norm numpy.exp loss.stable_kl self._dropout_p_map.get self.AttentionWrapper.super.__init__ json.loads new_sequence_outputs.append parse_args re.compile transformers.AutoTokenizer.from_pretrained.tokenize prefix.format.opt.get.lower x_flat.self.FC.self.f.self.linear.view plt.ylabel docs.append torch.float32.input.detach.F.softmax.sqrt_ ry.rp.p.sum data_config list v.contiguous.view pytorch_pretrained_bert.modeling.BertEmbeddings idx.mask.view apply_no_ans_threshold self.adv_teacher.forward module.dropout_wrapper.DropoutWrapper v.cpu config.get dataset.split idx.all_encoder_layers.detach train_config.add_argument self.FC sample_id_2_label_dic.keys weight.reshape proj.unsqueeze.x.bmm.squeeze self.SelfAttnWrapper.super.__init__ rng.randint data_utils.load_score_file numpy.exp.sum self.dropout_list.append data_utils.log_wrapper.create_logger torch.cuda.set_device dev_data_list.append query.transpose.size mt_dnn.inference.eval_model target.contiguous.view h0.size.h0.new.zero_ locals.items TASK_REGISTRY.get plt.step n_best_size.start_logits.np.argsort.tolist Vocabulary.add attn.transpose.transpose tokens.extend temp_2.pop experiments.ner.ner_utils.load_conll_pos self.Trilinear.super.__init__ logits.data.new numpy.ones_like p.nelement self.network.named_parameters pool.map numpy.power self.network.parameters rvl.append random.sample self.attn_list.append compare_output self.FlatSim.super.__init__ self.network.encode torch.bernoulli assert_dir_equal torch.save json.dump self.SimilarityWrapper.super.__init__ y_pred.append pooled_output.contiguous.view torch.distributed.init_process_group task_def.get args.task.lower embed.data.new vw.new_dict.cpu.numpy xWy.data.masked_fill_ self.DotProductProject.super.__init__ next batch_meta.batch_data.detach.cpu MaskedLmInstance self.num_hid.size.size.tmp_output.view.max.view Classifier x2.x1.abs score.np.argmax.tolist predict.strip.reshape self.reset query.transpose.transpose DotProductProject self._setup_adv_lossmap vb.new_dict.cpu generate_noise torch.cuda.device_count experiments.mlm.mlm_utils.load_loose_json.append batch_meta.batch_data.cuda torch.optim.SGD model.state_dict.items logit.size.logit.view.float self.x_linear rng.shuffle task_id.self.dropout_list.size sum tensorboardX.SummaryWriter os.access sys.path.append encoding_0.size.input_length.torch.LongTensor.fill_ x.size.x.size.x.data.new.zero_ sigmoid uid.gold_map.append x2_flat.self.proj_2.self.f.view rng.random golds.append weight.cpu.numpy tokenizer argparse.ArgumentParser WeightNorm p.size.p.contiguous.view.norm task_def_list.append self.qkv_head_dim.self.num_heads.bsz.src_len.k.contiguous.view.transpose.size experiments.exp_def.TaskDefs self._setup_adv_training line.strip.startswith self.optimizer.step self.adv_loss.update logit.size.logit.view.float.view args.model.split TrainingInstance torch.ones config_class.from_pretrained flat_scores.contiguous.view collections.namedtuple torch.log scipy.stats.pearsonr logging.getLogger.info grounds.append x1.size.x1.size.x1.contiguous.view.self.x_dot_linear.view.expand_as lang_map.items tasks.get_task_obj.input_is_valid_sample attn.transpose.contiguous self.att torch.optim.AdamW mt_dnn.perturbation.SmartPerturbation doc.split data_utils.tokenizer_utils.create_tokenizer._convert_token_to_id tokenizer.append embed.size.embed.data.new.normal_ torch.load numpy.concatenate p.dim torch.nn.functional.softmax build_data y.unsqueeze.expand_as self.Classifier.super.__init__ SimilarityWrapper logits.size.logits.data.new.zero_ attn.transpose.contiguous.view attn_weights.size.attn_weights.data.new.zero_ uid.predict_map.append key_padding_mask.unsqueeze.expand_as self._layer_norm self.EMA.super.__init__ start_position.append p.contiguous.view ent_strs.append torch.BoolTensor idx.lines.strip.split multiprocessing.Pool load_qnli torch.cuda.is_available sklearn.metrics.accuracy_score x.alpha.unsqueeze.bmm.squeeze logging.getLogger lower mask.size.score.np.argmax.reshape.tolist score.numpy.reshape group.setdefault mt_dnn.batcher.DistSingleTaskBatchSampler load_wnli human.items self.decoder_opt.append y.x.y.x.torch.cat.contiguous os.path.splitext pred.items predictions.append random.Random self.rebatch apex.amp.master_params numpy.ma.masked_array.mean mt_dnn.model.MTDNNModel.extract end.numpy.tolist load_scitail numpy.random.choice test_encoder tokenizer.pop attn_weights.float.F.softmax.type_as MLPSelfAttn data_utils.vocab.Vocabulary.add compute writer.write self.linear.unsqueeze tok_len.batch_size.torch.LongTensor.fill_ self.mnetwork model.to.state_dict end.data.cpu mt_dnn.inference.extract_encoding WeightNorm.compute_weight sklearn.metrics.f1_score target.contiguous.view.float instances.append output_file.open.read load_squad target.contiguous.view.detach experiments.glue.glue_utils.submit attention_mask_list.append model.encode load_qnnli self.nsp self.BilinearFlatSim.super.__init__ tqdm.auto.tqdm proj self.AdamaxW.super.__setstate__ mlm_y.view.view hyp_mask.size expected_dir.os.path.join.open.read masked_lm_labels.append p.contiguous epsilon.epsilon.y.detach.log test_metrics.items idx.all_encoder_layers.detach.cpu.numpy sub_part.pin_memory epsilon.epsilon.p.detach data_utils.utils.set_environment cand_indexes.append self._to_cuda len data_utils.utils.AverageMeter hid_shape.weight.new.zero_ apex.amp.scale_loss torch.no_grad torch.nn.functional.dropout sklearn.metrics.matthews_corrcoef model_class make_qid_to_has_ans remove_articles x1.size.x1.contiguous.view.self.x_linear.view.expand_as test_data_list.append vb.new_dict.cpu.numpy self.self_att task_id.self.dropout_list.contiguous x2.size.x2.contiguous.view.self.y_linear.view suffix.split.upper self.encoder_type.EncoderModelType.name.lower self.init_hidden logging.getLogger.setLevel self.tokenizer.batch_decode weight.size os.path.split self.scheduler.step input.view.float recalls.append torch.zeros self.MaskLmHeader.super.__init__ SanEncoder p.data.mul_ start.numpy.tolist current_chunk.append random.seed p.size.p.contiguous.view.norm.view filecmp.cmp sorted.insert eval.split load_rte functools.partial self._rnn.flatten_parameters model.predict v.cuda.cuda self._get_shuffled_index_batches_bin module.similarity.FlatSimilarityWrapper target.contiguous.view.size score.numpy.contiguous build_data_single positives.random.sample.pop qa_entry.get torch.ones_like torch.distributed.is_initialized kw.new_dict.cpu random.choice generate_decoder_opt encoding_0.size.input_lengths.max.input_lengths.len.torch.LongTensor.fill_ DotProduct torch.bmm mask.unsqueeze.expand_as format.replace train_config module.pooler.Pooler p.data.addcdiv_ self._get_shuffled_index_batches self.alpha.expand_as attn_weights.size.attn_weights.data.new.zero_.torch.diag.byte.unsqueeze load_boolq metric_max_over_ground_truths cs.LOSS_REGISTRY mask.sum self._get_batch_size getattr.cuda self.optimizer.state_dict argparse.ArgumentParser.parse_args opt.get attn_weights.size.attn_weights.data.new.zero_.torch.diag.byte.unsqueeze.expand_as load_cb bin_idx.data.append grad.abs dict argparse.ArgumentParser.add_argument exp_inf.mul_ len.pop data_utils.utils_qa.postprocess_qa_predictions x.data.new yaml.safe_load torch.nn.functional.softmax.unsqueeze WeightNorm.apply transformers.get_constant_schedule_with_warmup experiments.ner.ner_utils.load_conll_ner _mg RuntimeError noise.detach.detach label_tokenize prepare_train_feature SanLayer common.init_wrapper args.layers.split temp_2.append self.dropout.float plt.fill_between next.extend pytorch_pretrained_bert.modeling.BertLayerNorm n_best_size.end_logits.np.argsort.tolist self.task_loss_criterion.append float.key_padding_mask.unsqueeze.unsqueeze.attn_weights.float.masked_fill.type_as feature_extractor part.pin_memory.to load_cola input.view.detach x1.size.x1.contiguous.view.self.x_linear.view segment_ids.append sklearn.metrics.confusion_matrix tokens_a.extend experiments.mlm.mlm_utils.load_loose_json torch.cuda.manual_seed_all feature_index.features.get Wy.unsqueeze.x.bmm.squeeze str encoding_1.encoding_0.abs self.score_func.view torch.std y.x.torch.cat.contiguous self._task_type_map.keys embed.detach.abs task_id.self.dropout_list exact_scores.values seqeval.metrics.f1_score apex.amp.initialize plt.ylim torch.nn.parameter.Parameter blocks.lang_dict.append self.dense self.bert.parameters torch.nn.Parameter os.makedirs entry.get random.shuffle BilinearSum tasks.get_task_obj.test_prepare_label experiments.exp_def.TaskDef.from_dict line.strip.split.split self.tok2ind.get x1_flat.self.proj_1.self.f.view f1_scores.values sample_id_2_label_dic.items generate_golds_predictions_scores context.strip re.sub transformers.AutoTokenizer.from_pretrained.convert_tokens_to_ids self.config.get args.model.split.split mask.sum.tolist torch.nn.DataParallel input.view.squeeze transformers.AutoTokenizer.from_pretrained Trilinear register_task x2.contiguous.view self.network.load_state_dict module.register_parameter target.contiguous.view.contiguous experiments.exp_def.TaskDefs.get_task_def torch.FloatTensor tensorboardX.SummaryWriter.add_scalar sizes.append self.rnn_type.nn.getattr module.san.SANClassifier metric_func self.ind2tok.get self._setup_optim re.compile.findall kw.new_dict.cpu.numpy score_path.open.read prelim_predictions.append mt_dnn.matcher.SANBertNetwork glob.glob arch.config_class.from_pretrained.to_dict torch.bernoulli.unsqueeze Bilinear self.__random_select__ torch.distributed.get_rank self.model.named_parameters data_utils.roberta_utils.patch_name_dict model input.view.view line.strip.strip max_answer_seq_len.batch_size.torch.LongTensor.fill_ mt_dnn.batcher.MultiTaskDataset collections.Counter.values getattr math.sqrt logits.data.masked_fill_ qa_sample ImportError args.transformer_cache.init_model.config_class.from_pretrained.to_dict examples.append x2.size.x2.contiguous.view.self.y_linear.view.expand_as self.bert.embeddings data_utils.log_wrapper.create_logger.info experiments.mlm.mlm_utils.create_instances_from_document eff_perturb.item white_space_fix reduce_features_to_examples end_scores.squeeze.squeeze compute_acc feature_index.features.get.get experiments.ner.ner_utils.load_conll_chunk next.new predict.strip.strip exp_inf.new.long new_data.zero_.zero_ eff_noise.detach.abs.mean torch.stack json.load evaluation tasks.get_task_obj.train_prepare_soft_labels mt_dnn.batcher.SingleTaskDataset idx.all_encoder_layers.detach.cpu target.F.log_softmax.exp task_id.self.dropout_list.view self.config.get.item flat_squad load_wic_mtdnn batch_meta.batch_data.detach.cpu.numpy attn.transpose.size zip self.LayerNorm.super.__init__ task_layer p.size bais.cpu tokenizer.sequence_ids compute_exact os.path.join precisions.append y_true.append eps.grad.abs.add_.unsqueeze_ mlm_p.view.view repr model.cuda tuple x.contiguous.view print numpy.argsort load_snli model.size self.scoring_list.append
@developer
Could please help me check this issue?
May I pull a request to fix it?
Thank you very much.