Блог

Я смог обучить набор данных midi на основе архитектуры perofrmance_rnn_compact с помощью colab. Я также могу генерировать последовательность заметок с помощью:

 !performance_rnn_generate  --run_dir=/content/drive/MyDrive/Performance_RNN/training  --output_dir=/content/drive/MyDrive/Performance_RNN/generate  --config='performance_with_dynamics_compact'  --hparams="rnn_layer_sizes=[512,512,512]"  --temperature=1.  --num_outputs=5  --num_steps=20000  --primer_melody="[48,53,55,60,65,72,37]"  

к сожалению, я не могу создать пакет mag из моих обученных контрольных точек с помощью:

 #Save your model  !performance_rnn_generate  --run_dir=/content/drive/MyDrive/Performance_RNN/training  --hparams="rnn_layer_sizes=[512,512,512]"  --bundle_file=/content/drive/MyDrive/Performance_RNN/training/bundle.mag  --config='performance_with_dynamics_compact'  --save_generator_bundle  

Похоже, что полученный файл пакета может быть поврежден или работать неправильно. Он не может быть использован для создания новой последовательности midi-нот из него с помощью:

 !performance_rnn_generate  --config='performance_with_dynamics_compact'  --hparams="rnn_layer_sizes=[512,512,512]"  --bundle_file=/content/drive/MyDrive/Performance_RNN/training/bundle.mag  --output_dir=/content/drive/MyDrive/Performance_RNN/generate/outofbundle  --temperature=1.  --num_outputs=5  --num_steps=20000  --primer_melody="[48,53,55,60,65,72,37]"   

Я думаю, что это может быть как-то связано с «компактной» формой моей модели performance_with_dynamics_компактной модели?

Может быть, у кого-то есть идея?

Я получил следующую ошибку:

 /usr/local/lib/python3.7/dist-packages/librosa/util/decorators.py:9: NumbaDeprecationWarning: An import was requested from a module that has moved location. Import 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 /usr/local/lib/python3.7/dist-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 WARNING:tensorflow:From /usr/local/lib/python3.7/dist-packages/tensorflow/python/compat/v2_compat.py:111: 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 2021-11-15 15:22:02.008172: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0. INFO:tensorflow:Restoring parameters from /tmp/tmp81426j4g/model.ckpt I1115 15:22:02.059761 139764869441408 saver.py:1399] Restoring parameters from /tmp/tmp81426j4g/model.ckpt INFO:tensorflow:Need to generate 19912 more steps for this sequence, will try asking for 11948 RNN steps I1115 15:22:02.129988 139764869441408 performance_sequence_generator.py:240] Need to generate 19912 more steps for this sequence, will try asking for 11948 RNN steps Traceback (most recent call last):  File "/usr/local/bin/performance_rnn_generate", line 8, in lt;modulegt;  sys.exit(console_entry_point())  File "/usr/local/lib/python3.7/dist-packages/magenta/models/performance_rnn/performance_rnn_generate.py", line 290, in console_entry_point  tf.app.run(main)  File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/platform/app.py", line 40, in run  _run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)  File "/usr/local/lib/python3.7/dist-packages/absl/app.py", line 303, in run  _run_main(main, args)  File "/usr/local/lib/python3.7/dist-packages/absl/app.py", line 251, in _run_main  sys.exit(main(argv))  File "/usr/local/lib/python3.7/dist-packages/magenta/models/performance_rnn/performance_rnn_generate.py", line 285, in main  run_with_flags(generator)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/performance_rnn/performance_rnn_generate.py", line 244, in run_with_flags  generated_sequence = generator.generate(primer_sequence, generator_options)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/shared/sequence_generator.py", line 194, in generate  return self._generate(input_sequence, generator_options)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/performance_rnn/performance_sequence_generator.py", line 242, in _generate  len(performance)   rnn_steps_to_gen, performance, **args)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/performance_rnn/performance_model.py", line 81, in generate_performance  extend_control_events_callback=extend_control_events_callback)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/shared/events_rnn_model.py", line 378, in _generate_events  steps_per_iteration=steps_per_iteration)  File "/usr/local/lib/python3.7/dist-packages/magenta/common/beam_search.py", line 130, in beam_search  beam_entries, generate_step_fn, branch_factor, first_iteration_num_steps)  File "/usr/local/lib/python3.7/dist-packages/magenta/common/beam_search.py", line 63, in _generate_branches  all_sequences, all_states, all_scores)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/shared/events_rnn_model.py", line 232, in _generate_step  temperature)  File "/usr/local/lib/python3.7/dist-packages/magenta/models/shared/events_rnn_model.py", line 127, in _generate_step_for_batch  [graph_final_state, graph_softmax], feed_dict)  File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/client/session.py", line 971, in run  run_metadata_ptr)  File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/client/session.py", line 1168, in _run  f'Cannot feed value of shape {str(np_val.shape)} for Tensor ' ValueError: Cannot feed value of shape (1, 22, 388) for Tensor Placeholder:0, which has shape (1, ?, 1)  

Заранее спасибо, Кристиан