I'm using Python 3.6, latest version of Magenta and Tensorflow 1.15.2
(magenta-n) sven:~$ drums_rnn_train --config=drum_kit --run_dir=/tmp/drums_rnn/logdir/run1 --sequence_example_file=/tmp/drums_rnn/sequence_examples/train_drum_tracks.tfrecord --hparams="batch_size=5,rnn_layer_sizes=[64,64],attn_length=64" --num_training_steps=20000
/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/librosa/util/decorators.py:9: NumbaDeprecationWarning: An import was requested from a module that has moved location.
Import of 'jit' requested from: 'numba.decorators', please update to use 'numba.core.decorators' or pin to Numba version 0.48.0. This alias will not be present in Numba version 0.50.0.
from numba.decorators import jit as optional_jit
/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/pydub/utils.py:170: RuntimeWarning: Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work
warn("Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work", RuntimeWarning)
WARNING:tensorflow:From /home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/tensorflow_core/python/compat/v2_compat.py:68: disable_resource_variables (from tensorflow.python.ops.variable_scope) is deprecated and will be removed in a future version.
Instructions for updating:
non-resource variables are not supported in the long term
INFO:tensorflow:hparams = {'batch_size': 5, 'rnn_layer_sizes': [64, 64], 'dropout_keep_prob': 0.5, 'attn_length': 64, 'clip_norm': 3, 'learning_rate': 0.001, 'residual_connections': False, 'use_cudnn': False}
I1122 17:14:07.142286 139790753481088 events_rnn_graph.py:96] hparams = {'batch_size': 5, 'rnn_layer_sizes': [64, 64], 'dropout_keep_prob': 0.5, 'attn_length': 64, 'clip_norm': 3, 'learning_rate': 0.001, 'residual_connections': False, 'use_cudnn': False}
INFO:tensorflow:Train dir: /tmp/drums_rnn/logdir/run1/train
I1122 17:14:07.142590 139790753481088 drums_rnn_train.py:87] Train dir: /tmp/drums_rnn/logdir/run1/train
WARNING:tensorflow:From /home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/common/sequence_example_lib.py:66: string_input_producer (from tensorflow.python.training.input) is deprecated and will be removed in a future version.
Instructions for updating:
Queue-based input pipelines have been replaced by `tf.data`. Use `tf.data.Dataset.from_tensor_slices(string_tensor).shuffle(tf.shape(input_tensor, out_type=tf.int64)[0]).repeat(num_epochs)`. If `shuffle=False`, omit the `.shuffle(...)`.
W1122 17:14:07.143937 139790753481088 deprecation.py:323] From /home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/common/sequence_example_lib.py:66: string_input_producer (from tensorflow.python.training.input) is deprecated and will be removed in a future version.
Instructions for updating:
Queue-based input pipelines have been replaced by `tf.data`. Use `tf.data.Dataset.from_tensor_slices(string_tensor).shuffle(tf.shape(input_tensor, out_type=tf.int64)[0]).repeat(num_epochs)`. If `shuffle=False`, omit the `.shuffle(...)`.
Traceback (most recent call last):
File "/home/sven/miniconda3/envs/magenta-n/bin/drums_rnn_train", line 8, in <module>
sys.exit(console_entry_point())
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/models/drums_rnn/drums_rnn_train.py", line 109, in console_entry_point
tf.app.run(main)
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/tensorflow_core/python/platform/app.py", line 40, in run
_run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/absl/app.py", line 303, in run
_run_main(main, args)
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/absl/app.py", line 251, in _run_main
sys.exit(main(argv))
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/models/drums_rnn/drums_rnn_train.py", line 104, in main
checkpoints_to_keep=FLAGS.num_checkpoints)
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/models/shared/events_rnn_train.py", line 54, in run_training
build_graph_fn()
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/models/shared/events_rnn_graph.py", line 113, in build
label_shape=label_shape, shuffle=mode == 'train')
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/magenta/common/sequence_example_lib.py", line 66, in get_padded_batch
file_queue = tf.train.string_input_producer(file_list)
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/tensorflow_core/python/util/deprecation.py", line 324, in new_func
return func(*args, **kwargs)
File "/home/sven/miniconda3/envs/magenta-n/lib/python3.6/site-packages/tensorflow_core/python/training/input.py", line 258, in string_input_producer
raise ValueError(not_null_err)
ValueError: string_input_producer requires a non-null input tensor
Thanks a lot. My training data is correctly created:
DAGPipeline_DrumsExtractor_eval_drum_tracks_truncated: 5
INFO:tensorflow:DAGPipeline_DrumsExtractor_training_drum_track_lengths_in_bars:
[7,8): 1
[10,20): 1
[20,30): 3
[30,40): 8
I1122 17:12:02.001522 140077954343296 statistics.py:137] DAGPipeline_DrumsExtractor_training_drum_track_lengths_in_bars:
[7,8): 1
[10,20): 1
[20,30): 3
[30,40): 8
I've tried almost everything, create a new conda environment, changed the command, changes my training set etc. nothing helped.
