Jasper can't find saved model

[akafen]

hi, thank you developed this great toolkit and open source.but i have some problem and cant solve by myself.

i run train.py in examples/japer.After 6 epoches, loss reduce to 0.69. But I don't find any saved model in my saved directory

i think i was follow the instruction in readme.

python version:3.7.3 tf version: 2.3.2 ctc_loss in tf use characters

config.yml:

speech_config: sample_rate: 16000 frame_ms: 25 stride_ms: 10 num_feature_bins: 80 feature_type: log_mel_spectrogram preemphasis: 0.97 normalize_signal: True normalize_feature: True normalize_per_feature: False

decoder_config: vocabulary: null blank_at_zero: False beam_width: 500 lm_config: model_path: null alpha: 2.0 beta: 1.0

model_config: name: jasper dense: True first_additional_block_channels: 256 first_additional_block_kernels: 11 first_additional_block_strides: 2 first_additional_block_dilation: 1 first_additional_block_dropout: 0.2 nsubblocks: 3 block_channels: [256, 384, 512, 640, 768] block_kernels: [11, 13, 17, 21, 25] block_dropout: [0.2, 0.2, 0.2, 0.3, 0.3] second_additional_block_channels: 896 second_additional_block_kernels: 1 second_additional_block_strides: 1 second_additional_block_dilation: 2 second_additional_block_dropout: 0.4 third_additional_block_channels: 1024 third_additional_block_kernels: 1 third_additional_block_strides: 1 third_additional_block_dilation: 1 third_additional_block_dropout: 0.4

learning_config: train_dataset_config: use_tf: True data_paths: - /home/liuyijiao/tensorflow_asr_datasets/tensorflow_asr.tsv tfrecords_dir: /home/liuyijiao/tensorflow_asr_datasets/tfrecords shuffle: True cache: True buffer_size: 128 drop_remainder: True stage: train

eval_dataset_config: use_tf: True data_paths: - /home/liuyijiao/tensorflow_asr_datasets/valid_transcripts.tsv tfrecords_dir: /home/liuyijiao/tensorflow_asr_datasets/tfrecords shuffle: False cache: True buffer_size: 100 drop_remainder: True stage: eval

test_dataset_config: use_tf: True data_paths: - /home/liuyijiao/tensorflow_asr_datasets/test_transcripts.tsv tfrecords_dir: /home/liuyijiao/tensorflow_asr_datasets/tfrecords shuffle: False cache: True buffer_size: 100 drop_remainder: True stage: test

optimizer_config: class_name: adam config: learning_rate: 0.0001

running_config: batch_size: 32 num_epochs: 6 checkpoint: filepath: /home/liuyijiao/tensorflow_asr_datasets/models/jasper/checkpoints/{epoch:02d}.h5 save_best_only: True save_weights_only: False save_freq: epoch states_dir: /home/liuyijiao/tensorflow_asr_datasets/models/jasper/states tensorboard: log_dir: /home/liuyijiao/tensorflow_asr_datasets/models/jasper/tensorboard histogram_freq: 1 write_graph: True write_images: True update_freq: epoch profile_batch: 2

[nglehuy]

@akafen Try setting save_best_only: False in the checkpoint option to make sure it saves every check points

[nglehuy]

I’ll close the issue here due to inactivity. Feel free to reopen if you have further questions.