Getting a graph execution error / JIT compilation failed with latest versions

[bluenote10]

Using the latest versions of

• crepe==0.0.14
• tensorflow==2.15.0.post1

I'm getting a graph execution error / JIT compilation failed with the following reproduction code:

import crepe
import numpy as np

sr = 16000
step_size = 4

signal = np.random.normal(0.0, 0.1, size=sr)

times, f0, f0_conf, _ = crepe.predict(
    signal,
    sr,
    step_size=step_size,
    verbose=1,
)

The full output is somewhat lengthy:

2024-01-08 22:51:49.982191: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2024-01-08 22:51:49.982227: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2024-01-08 22:51:49.983128: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-01-08 22:51:49.988289: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-01-08 22:51:50.702792: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
2024-01-08 22:51:51.233335: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.279192: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.279367: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.279894: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.280020: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.280132: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.323194: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.323360: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.323500: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-01-08 22:51:51.323593: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1929] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 2881 MB memory:  -> device: 0, name: NVIDIA GeForce GTX 980, pci bus id: 0000:01:00.0, compute capability: 5.2
2024-01-08 22:51:51.947998: I external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:454] Loaded cuDNN version 8902
2024-01-08 22:51:52.719045: W external/local_xla/xla/service/gpu/llvm_gpu_backend/gpu_backend_lib.cc:504] Can't find libdevice directory ${CUDA_DIR}/nvvm/libdevice. This may result in compilation or runtime failures, if the program we try to run uses routines from libdevice.
Searched for CUDA in the following directories:
  ./cuda_sdk_lib
  /usr/local/cuda-12.2
  /usr/local/cuda
  /home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/tensorflow/python/platform/../../../nvidia/cuda_nvcc
  /home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/tensorflow/python/platform/../../../../nvidia/cuda_nvcc
  .
You can choose the search directory by setting xla_gpu_cuda_data_dir in HloModule's DebugOptions.  For most apps, setting the environment variable XLA_FLAGS=--xla_gpu_cuda_data_dir=/path/to/cuda will work.
2024-01-08 22:51:53.513126: W external/local_tsl/tsl/framework/bfc_allocator.cc:296] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.59GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available.
2024-01-08 22:51:53.529700: W external/local_tsl/tsl/framework/bfc_allocator.cc:296] Allocator (GPU_0_bfc) ran out of memory trying to allocate 4.59GiB with freed_by_count=0. The caller indicates that this is not a failure, but this may mean that there could be performance gains if more memory were available.
2024-01-08 22:51:53.607953: W external/local_xla/xla/service/gpu/llvm_gpu_backend/gpu_backend_lib.cc:542] libdevice is required by this HLO module but was not found at ./libdevice.10.bc
error: libdevice not found at ./libdevice.10.bc
2024-01-08 22:51:53.608118: E tensorflow/compiler/mlir/tools/kernel_gen/tf_framework_c_interface.cc:207] INTERNAL: Generating device code failed.
2024-01-08 22:51:53.608752: W tensorflow/core/framework/op_kernel.cc:1827] UNKNOWN: JIT compilation failed.
Traceback (most recent call last):
  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/home/me/git/AudioML/audio_ml/experiments/debug/debug_crepe.py", line 13, in <module>
    times, f0, f0_conf, _ = crepe.predict(
  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/crepe/core.py", line 255, in predict
    activation = get_activation(audio, sr, model_capacity=model_capacity,
  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/crepe/core.py", line 212, in get_activation
    return model.predict(frames, verbose=verbose)
  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 70, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/tensorflow/python/eager/execute.py", line 53, in quick_execute
    tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
tensorflow.python.framework.errors_impl.UnknownError: Graph execution error:

Detected at node model/classifier/Sigmoid defined at (most recent call last):
  File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main

  File "/usr/lib/python3.10/runpy.py", line 86, in _run_code

  File "/home/me/git/AudioML/audio_ml/experiments/debug/debug_crepe.py", line 13, in <module>

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/crepe/core.py", line 255, in predict

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/crepe/core.py", line 212, in get_activation

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 65, in error_handler

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/training.py", line 2655, in predict

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/training.py", line 2440, in predict_function

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/training.py", line 2425, in step_function

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/training.py", line 2413, in run_step

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/training.py", line 2381, in predict_step

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 65, in error_handler

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/training.py", line 590, in __call__

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 65, in error_handler

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/base_layer.py", line 1149, in __call__

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 96, in error_handler

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/functional.py", line 515, in call

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/functional.py", line 672, in _run_internal_graph

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 65, in error_handler

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/engine/base_layer.py", line 1149, in __call__

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py", line 96, in error_handler

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/layers/core/dense.py", line 255, in call

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/activations.py", line 400, in sigmoid

  File "/home/me/.virtualenvs/ddsp_pytorch/lib/python3.10/site-packages/keras/src/backend.py", line 5915, in sigmoid

JIT compilation failed.
	 [[{{node model/classifier/Sigmoid}}]] [Op:__inference_predict_function_759]

crepe used to work fine before on my machine with older versions so this may be a regression.