search

fundamentalvision / Uni-Perceiver

Apache License 2.0
256 stars 21 forks source link

在GLUE数据集上测试，程序报错 #11

Closed tgyy1995 closed 1 year ago

tgyy1995 commented 1 year ago

CoLA:

GLUE_CoLA.yaml

```yaml _BASE_: "base.yaml" SHARED_TARGETS: - NAME: 'CoLA' SHARED_TARGETS_CFG: FILE_PATH: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/GLUE_classnames/CoLA_class_name_CLIP_with_endoftext.pkl' DISTRIBUTED: False TASKS: - NAME: CoLA DATASETS: TRAIN: 'GLUEDataset' TEST: 'GLUEDataset' #VAL: 'GLUEDataset' TASK_TYPE: 'text_classification' DATASET_NAME: 'CoLA' TARGET_SET: ['CoLA'] DATALOADER: TRAIN_BATCH_SIZE: 16 TEST_BATCH_SIZE: 64 NUM_WORKERS: 4 ANNO_FOLDER: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/bert_pretrain_data/glue_data/' MODEL: MAX_SEQ_LEN: 256 TEMP_NAME: logit_scale_text_mlm LOSSES: NAMES: ['LabelSmoothingCrossEntropy', 'Accuracy'] LABELSMOOTHING: 0.1 # LOSS_WEIGHT: 1 REDUCTION: 'mean' LOSS_FP32: False INFERENCE: NAME: 'GLUEEvaler' VOCAB: 'CLIP' GENERATION_MODE: False ENGINE: NAME: 'UnifiedTrainer' DATALOADER: USE_WEIGHTED_SAMPLER: True UNIFIED_DATASET: True NUM_WORKERS: 16 ######################################### MODEL ######################################### MODEL: MODEL_EMA: False MODEL_EMA_DECAY: 0.9999 ####################################### Optimizer ####################################### SOLVER: NAME: 'Adam' # EPOCH: 1 MAX_ITER: 5600 CHECKPOINT_PERIOD: 1000000 EVAL_PERIOD: 200 CHECKPOINT_MAX_SAVE: 1 BASE_LR: 0.00001 BIAS_LR_FACTOR: 1.0 WEIGHT_DECAY: 0.1 WEIGHT_DECAY_NORM: 0.0 WEIGHT_DECAY_BIAS: 0.0 MOMENTUM: 0.9 DAMPENING: 0.0 NESTEROV: 0.0 BETAS: [0.9, 0.98] EPS: 1e-8 GRAD_CLIP: 0.5 GRAD_CLIP_TYPE: 'norm' ACCUM_ITER: 0 AMP_FP16: True APEX_FP16: False # dangerous WRITE_PERIOD: 20 ####################################### lr scheduler ####################################### LR_SCHEDULER: NAME: 'WarmupCosine' WARMUP: 400 MIN_LR: 0.00000001 find_unused_parameters: true ```

error

``` Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu)engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data datalist = self.load_raw_data(cfg)init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 225, in load_raw_data File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data label = float(sensinfo[self.label_column - 1]) ValueError: could not convert string to float: 'Bill whistled past the house.' datalist = self.load_raw_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 225, in load_raw_data label = float(sensinfo[self.label_column - 1]) ValueError: could not convert string to float: 'Bill whistled past the house.' ```

#######################################################

GLUE_MNLI.yaml

```yaml _BASE_: "base.yaml" SHARED_TARGETS: - NAME: 'MNLI' SHARED_TARGETS_CFG: FILE_PATH: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/GLUE_classnames/MNLI_class_name_CLIP_with_endoftext.pkl' DISTRIBUTED: False TASKS: - NAME: MNLI DATASETS: TRAIN: 'GLUEDataset' TEST: 'GLUEDataset' #VAL: 'GLUEDataset' TASK_TYPE: 'text_classification' DATASET_NAME: 'MNLI_Match' TARGET_SET: ['MNLI'] DATALOADER: TRAIN_BATCH_SIZE: 32 TEST_BATCH_SIZE: 32 NUM_WORKERS: 4 ANNO_FOLDER: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/bert_pretrain_data/glue_data/' MODEL: MAX_SEQ_LEN: 256 TEMP_NAME: logit_scale_text_mlm LOSSES: NAMES: ['LabelSmoothingCrossEntropy', 'Accuracy'] LABELSMOOTHING: 0.1 # LOSS_WEIGHT: 1 REDUCTION: 'mean' LOSS_FP32: False INFERENCE: NAME: 'GLUEEvaler' VOCAB: 'CLIP' GENERATION_MODE: False ENGINE: NAME: 'UnifiedTrainer' DATALOADER: USE_WEIGHTED_SAMPLER: True UNIFIED_DATASET: True NUM_WORKERS: 16 ######################################### MODEL ######################################### MODEL: MODEL_EMA: False MODEL_EMA_DECAY: 0.9999 ####################################### Optimizer ####################################### SOLVER: NAME: 'Adam' MAX_ITER: 125000 CHECKPOINT_PERIOD: 125000 EVAL_PERIOD: 5000 CHECKPOINT_MAX_SAVE: 1 BASE_LR: 0.00001 BIAS_LR_FACTOR: 1.0 WEIGHT_DECAY: 0.1 WEIGHT_DECAY_NORM: 0.0 WEIGHT_DECAY_BIAS: 0.0 MOMENTUM: 0.9 DAMPENING: 0.0 NESTEROV: 0.0 BETAS: [0.9, 0.98] EPS: 1e-8 GRAD_CLIP: 0.5 GRAD_CLIP_TYPE: 'norm' ACCUM_ITER: 0 AMP_FP16: True APEX_FP16: False # dangerous WRITE_PERIOD: 20 ####################################### lr scheduler ####################################### LR_SCHEDULER: NAME: 'WarmupCosine' WARMUP: 7500 MIN_LR: 0.00000001 find_unused_parameters: true ```

error

``` Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data datalist = self.load_raw_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 219, in load_raw_data label = labelmapping[sensinfo[self.label_column - 1]] IndexError: list index out of range Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data datalist = self.load_raw_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 219, in load_raw_data label = labelmapping[sensinfo[self.label_column - 1]] IndexError: list index out of range ```

##################################################

GLUE_MRPC.yaml

```yaml _BASE_: "base.yaml" SHARED_TARGETS: - NAME: 'MRPC' SHARED_TARGETS_CFG: FILE_PATH: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/GLUE_classnames/MRPC_class_name_CLIP_with_endoftext.pkl' DISTRIBUTED: False TASKS: - NAME: MRPC DATASETS: TRAIN: 'GLUEDataset' TEST: 'GLUEDataset' #VAL: 'GLUEDataset' TASK_TYPE: 'text_classification' DATASET_NAME: 'MRPC' TARGET_SET: ['MRPC'] DATALOADER: TRAIN_BATCH_SIZE: 16 TEST_BATCH_SIZE: 64 NUM_WORKERS: 4 ANNO_FOLDER: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/bert_pretrain_data/glue_data/' MODEL: MAX_SEQ_LEN: 256 TEMP_NAME: logit_scale_text_mlm LOSSES: NAMES: ['LabelSmoothingCrossEntropy', 'Accuracy'] LABELSMOOTHING: 0.1 # LOSS_WEIGHT: 1 REDUCTION: 'mean' LOSS_FP32: False INFERENCE: NAME: 'GLUEEvaler' VOCAB: 'CLIP' GENERATION_MODE: False ENGINE: NAME: 'UnifiedTrainer' DATALOADER: USE_WEIGHTED_SAMPLER: True UNIFIED_DATASET: True NUM_WORKERS: 16 ######################################### MODEL ######################################### MODEL: MODEL_EMA: False MODEL_EMA_DECAY: 0.9999 ####################################### Optimizer ####################################### SOLVER: NAME: 'Adam' MAX_ITER: 2500 CHECKPOINT_PERIOD: 10000 EVAL_PERIOD: 100 CHECKPOINT_MAX_SAVE: 2 BASE_LR: 0.00001 BIAS_LR_FACTOR: 1.0 WEIGHT_DECAY: 0.1 WEIGHT_DECAY_NORM: 0.0 WEIGHT_DECAY_BIAS: 0.0 MOMENTUM: 0.9 DAMPENING: 0.0 NESTEROV: 0.0 BETAS: [0.9, 0.98] EPS: 1e-8 GRAD_CLIP: 0.5 GRAD_CLIP_TYPE: 'norm' ACCUM_ITER: 0 AMP_FP16: True APEX_FP16: False # dangerous WRITE_PERIOD: 20 ####################################### lr scheduler ####################################### LR_SCHEDULER: NAME: 'WarmupCosine' WARMUP: 150 MIN_LR: 0.00000001 find_unused_parameters: true ```

error

``` Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 144, in main res = trainer.test(trainer.cfg, trainer.model, trainer.test_data_loader, trainer.test_evaluator, epoch=-1) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 520, in test return tester(self.task_cfg, model, test_data_loader, evaluator, epoch, self.amp_fp16, self.apex_fp16) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_tester.py", line 92, in tester results[task] = test_cls(task_cfg[task], model, test_data_loader[task], evaluator[task], epoch, amp_fp16, task) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_tester.py", line 239, in test_cls eval_res = evaluator.eval(results, epoch) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/evaluation/glue_evaler.py", line 56, in eval f1 = f1_score(y_true=labels, y_pred=preds) File "/home/whl/anaconda3/envs/uni-perceiver2/lib/python3.9/site-packages/sklearn/metrics/_classification.py", line 1123, in f1_score return fbeta_score( File "/home/whl/anaconda3/envs/uni-perceiver2/lib/python3.9/site-packages/sklearn/metrics/_classification.py", line 1261, in fbeta_score _, _, f, _ = precision_recall_fscore_support( File "/home/whl/anaconda3/envs/uni-perceiver2/lib/python3.9/site-packages/sklearn/metrics/_classification.py", line 1544, in precision_recall_fscore_support labels = _check_set_wise_labels(y_true, y_pred, average, labels, pos_label) File "/home/whl/anaconda3/envs/uni-perceiver2/lib/python3.9/site-packages/sklearn/metrics/_classification.py", line 1365, in _check_set_wise_labels raise ValueError( ValueError: Target is multiclass but average='binary'. Please choose another average setting, one of [None, 'micro', 'macro', 'weighted'].```

###################################################### ##################################################

GLUE_QNLI.yaml

```yaml _BASE_: "base.yaml" SHARED_TARGETS: - NAME: 'QNLI' SHARED_TARGETS_CFG: FILE_PATH: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/GLUE_classnames/QNLI_class_name_CLIP_with_endoftext.pkl' DISTRIBUTED: False TASKS: - NAME: QNLI DATASETS: TRAIN: 'GLUEDataset' TEST: 'GLUEDataset' #VAL: 'GLUEDataset' TASK_TYPE: 'text_classification' DATASET_NAME: 'QNLI' TARGET_SET: ['QNLI'] DATALOADER: TRAIN_BATCH_SIZE: 16 TEST_BATCH_SIZE: 64 NUM_WORKERS: 4 ANNO_FOLDER: '/mnt/14T-disk/code/Uni-Perceiver5/datasets/bert_pretrain_data/glue_data/' MODEL: MAX_SEQ_LEN: 256 TEMP_NAME: logit_scale_text_mlm LOSSES: NAMES: ['LabelSmoothingCrossEntropy', 'Accuracy'] LABELSMOOTHING: 0.1 # LOSS_WEIGHT: 1 REDUCTION: 'mean' LOSS_FP32: False INFERENCE: NAME: 'GLUEEvaler' VOCAB: 'CLIP' GENERATION_MODE: False ENGINE: NAME: 'UnifiedTrainer' DATALOADER: USE_WEIGHTED_SAMPLER: True UNIFIED_DATASET: True NUM_WORKERS: 16 ######################################### MODEL ######################################### MODEL: MODEL_EMA: False MODEL_EMA_DECAY: 0.9999 ####################################### Optimizer ####################################### SOLVER: NAME: 'Adam' MAX_ITER: 34000 CHECKPOINT_PERIOD: 200000 EVAL_PERIOD: 2000 CHECKPOINT_MAX_SAVE: 2 BASE_LR: 0.00001 BIAS_LR_FACTOR: 1.0 WEIGHT_DECAY: 0.1 WEIGHT_DECAY_NORM: 0.0 WEIGHT_DECAY_BIAS: 0.0 MOMENTUM: 0.9 DAMPENING: 0.0 NESTEROV: 0.0 BETAS: [0.9, 0.98] EPS: 1e-8 GRAD_CLIP: 0.5 GRAD_CLIP_TYPE: 'norm' ACCUM_ITER: 0 AMP_FP16: True APEX_FP16: False # dangerous WRITE_PERIOD: 20 ####################################### lr scheduler ####################################### LR_SCHEDULER: NAME: 'WarmupCosine' WARMUP: 2000 MIN_LR: 0.00000001 find_unused_parameters: true ```

error

``` Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data datalist = self.load_raw_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 221, in load_raw_data label = 1.0 if sensinfo[self.label_column - 1] == "entailment" else 0.0 IndexError: list index out of range Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data datalist = self.load_raw_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 221, in load_raw_data label = 1.0 if sensinfo[self.label_column - 1] == "entailment" else 0.0 IndexError: list index out of range ```

###################################################### ################################################## QQP:

error

``` Traceback (most recent call last): Traceback (most recent call last): File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 171, in main(args)main(args) File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main File "/mnt/14T-disk/code/Uni-Perceiver5/main.py", line 113, in main trainer = build_engine(cfg)trainer = build_engine(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/build.py", line 10, in build_engine engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg)engine = ENGINE_REGISTRY.get(cfg.ENGINE.NAME)(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/engine/unified_trainer.py", line 117, in __init__ self.test_data_loader = build_test_loader(cfg, self.task_cfg)self.test_data_loader = build_test_loader(cfg, self.task_cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/utils/engine_util.py", line 74, in build_test_loader loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu)loaders[name] = build_standard_valtest_loader(new_cfg, task_cfg, stage='test', multi_gpu_eval=multi_gpu) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 327, in build_standard_valtest_loader dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage)dataset = build_dataset_mapper(cfg, name=dataset_name, stage=stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/build.py", line 41, in build_dataset_mapper dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage)dataset_mapper = DATASETS_REGISTRY.get(name)(cfg, stage) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/config/config.py", line 211, in wrapped init_func(self, **explicit_args)init_func(self, **explicit_args) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 59, in __init__ self.load_data(cfg)self.load_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 159, in load_data datalist = self.load_raw_data(cfg)datalist = self.load_raw_data(cfg) File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 225, in load_raw_data File "/mnt/14T-disk/code/Uni-Perceiver5/uniperceiver/datasets/task_dataset/GLUE.py", line 225, in load_raw_data label = float(sensinfo[self.label_column - 1])label = float(sensinfo[self.label_column - 1]) IndexErrorIndexError: : list index out of rangelist index out of range ```

Lechatelia commented 1 year ago

sorry for this error. I will check it as soon as possible.

Lechatelia commented 1 year ago

初步看来应该是glue数据的标注文件处理之后存在着差异，我们是参考了fairseq的前处理方式 这里有详细的介绍prepare_data.md

您方便将某一个任务的文件发给我吗？比如 /glue_data/CoLA/processed/train.tsv 可以发给我的邮箱 lechatelia@stu.xjtu.edu.cn

tgyy1995 commented 1 year ago

抱歉，我在问题里可能没有说清楚， 我在config中添加TASKS.DATASETS.TEST: 'GLUEdataset' 来在测试集上运行时，程序会出错。不添加 TASKS.DATASETS.TEST: 'GLUEdataset' 时程序是可以正常训练的。

------------------ 原始邮件 ------------------ 发件人: "fundamentalvision/Uni-Perceiver" @.>; 发送时间: 2023年2月14日(星期二) 晚上7:30 @.>; @.**@.>; 主题: Re: [fundamentalvision/Uni-Perceiver] 在GLUE数据集上测试，程序报错 (Issue #11)

初步看来应该是glue数据的标注文件处理之后存在着差异，我们是参考了fairseq的前处理方式这里有详细的介绍) （prepare_data.md

您方便将某一个任务的文件发给我吗？比如 /glue_data/CoLA/processed/train.tsv 可以发给我的邮箱 @.***

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

Lechatelia commented 1 year ago

不好意思 这份code没有验证测试集是否存在问题， 我会尽快验证下GLUE测试集涉及到的code是否存在问题，