Created by: liu824

训练时我遇到下面的问题： [INFO: train.py: 249]: Namespace(batch_size=None, config='models-1.6/PaddleCV/PaddleVideo/configs/bmn.yaml', epoch=None, fix_random_seed=False, learning_rate=None, log_interval=10, model_name='BMN', no_memory_optimize=False, pretrain=None, resume=None, save_dir='./data/checkpoints', use_gpu=True, valid_interval=1) [INFO: config_utils.py: 70]: ---------------- Train Arguments ---------------- [INFO: config_utils.py: 72]: MODEL: [INFO: config_utils.py: 74]: name:BMN [INFO: config_utils.py: 74]: tscale:551 [INFO: config_utils.py: 74]: dscale:551 [INFO: config_utils.py: 74]: feat_dim:1024 [INFO: config_utils.py: 74]: prop_boundary_ratio:0.5 [INFO: config_utils.py: 74]: num_sample:2 [INFO: config_utils.py: 74]: num_sample_perbin:3 [INFO: config_utils.py: 74]: anno_file:trainnew.json [INFO: config_utils.py: 74]: feat_path:work/trainnew/videofeature [INFO: config_utils.py: 72]: TRAIN: [INFO: config_utils.py: 74]: subset:train [INFO: config_utils.py: 74]: epoch:9 [INFO: config_utils.py: 74]: batch_size:2 [INFO: config_utils.py: 74]: num_threads:8 [INFO: config_utils.py: 74]: use_gpu:True [INFO: config_utils.py: 74]: num_gpus:4 [INFO: config_utils.py: 74]: learning_rate:0.001 [INFO: config_utils.py: 74]: learning_rate_decay:0.1 [INFO: config_utils.py: 74]: lr_decay_iter:4200 [INFO: config_utils.py: 74]: l2_weight_decay:0.0001 [INFO: config_utils.py: 72]: VALID: [INFO: config_utils.py: 74]: subset:validation [INFO: config_utils.py: 74]: batch_size:16 [INFO: config_utils.py: 74]: num_threads:8 [INFO: config_utils.py: 74]: use_gpu:True [INFO: config_utils.py: 74]: num_gpus:4 [INFO: config_utils.py: 72]: TEST: [INFO: config_utils.py: 74]: subset:validation [INFO: config_utils.py: 74]: batch_size:1 [INFO: config_utils.py: 74]: num_threads:1 [INFO: config_utils.py: 74]: snms_alpha:0.001 [INFO: config_utils.py: 74]: snms_t1:0.5 [INFO: config_utils.py: 74]: snms_t2:0.9 [INFO: config_utils.py: 74]: output_path:data/output/EVAL/BMN_results [INFO: config_utils.py: 74]: result_path:data/evaluate_results [INFO: config_utils.py: 72]: INFER: [INFO: config_utils.py: 74]: subset:test [INFO: config_utils.py: 74]: batch_size:1 [INFO: config_utils.py: 74]: num_threads:1 [INFO: config_utils.py: 74]: snms_alpha:0.4 [INFO: config_utils.py: 74]: snms_t1:0.5 [INFO: config_utils.py: 74]: snms_t2:0.9 [INFO: config_utils.py: 74]: filelist:data/dataset/bmn/infer.list [INFO: config_utils.py: 74]: output_path:data/output/INFER/BMN_results [INFO: config_utils.py: 74]: result_path:data/predict_results [INFO: config_utils.py: 75]: ------------------------------------------------- W0701 10:13:39.680063 368 device_context.cc:252] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 9.2, Runtime API Version: 9.0 W0701 10:13:39.683766 368 device_context.cc:260] device: 0, cuDNN Version: 7.3. train subset video numbers: 3500 validation subset video numbers: 0 [INFO: bmn_proposal_metrics.py: 62]: Resetting train metrics... [INFO: bmn_proposal_metrics.py: 62]: Resetting valid metrics... [INFO: train_utils.py: 45]: ------- learning rate [0.], learning rate counter [-1] ----- [INFO: train_utils.py: 104]: No iteration was executed, please check the data reader 我的配置文件如下： MODEL: name: "BMN" tscale: 551 dscale: 551 feat_dim: 1024 prop_boundary_ratio: 0.5 num_sample: 2 num_sample_perbin: 3 anno_file: "trainnew.json" feat_path: 'work/trainnew/videofeature'

TRAIN: subset: "train" epoch: 9 batch_size: 2 num_threads: 8 use_gpu: True num_gpus: 4 learning_rate: 0.001 learning_rate_decay: 0.1 lr_decay_iter: 4200 l2_weight_decay: 1e-4 其他没有变化