to cpu

batdetect2/data/iterators.py

@@ -0,0 +1,79 @@
+from collections.abc import Generator
+from typing import Optional, Tuple
+
+from soundevent import data
+
+from batdetect2.data.datasets import Dataset
+from batdetect2.targets.types import TargetProtocol
+
+
+def iterate_over_sound_events(
+    dataset: Dataset,
+    targets: TargetProtocol,
+    apply_filter: bool = True,
+    apply_transform: bool = True,
+    exclude_generic: bool = True,
+) -> Generator[Tuple[Optional[str], data.SoundEventAnnotation], None, None]:
+    """Iterate over sound events in a dataset, applying filtering and
+    transformations.
+
+    This generator function processes sound event annotations from a given
+    dataset, allowing for optional filtering, transformation, and exclusion of
+    unclassifiable (generic) events based on the provided target definitions.
+
+    Parameters
+    ----------
+    dataset : Dataset
+        The dataset containing clip annotations, each of which may contain
+        multiple sound event annotations.
+    targets : TargetProtocol
+        An object implementing the `TargetProtocol`, which provides methods
+        for filtering, transforming, and encoding sound events.
+    apply_filter : bool, optional
+        If True, sound events will be filtered using `targets.filter()`.
+        Only events for which `targets.filter()` returns True will be yielded.
+        Defaults to True.
+    apply_transform : bool, optional
+        If True, sound events will be transformed using `targets.transform()`
+        before being yielded. Defaults to True.
+    exclude_generic : bool, optional
+        If True, sound events that result in a `None` class name after
+        `targets.encode()` will be excluded. This is typically used to
+        filter out events that cannot be mapped to a specific target class.
+        Defaults to True.
+
+    Yields
+    ------
+    Tuple[Optional[str], data.SoundEventAnnotation]
+        A tuple containing:
+        - The encoded class name (str) for the sound event, or None if it
+          cannot be encoded to a specific class.
+        - The sound event annotation itself, after passing all specified
+          filtering and transformation steps.
+
+    Notes
+    -----
+    The processing order for each sound event is:
+    1. Filtering (if `apply_filter` is True). Events failing the filter are
+       skipped.
+    2. Transformation (if `apply_transform` is True).
+    3. Encoding to determine class name and check for genericity (if
+       `exclude_generic` is True). Events with a `None` class name are skipped
+        if `exclude_generic` is True.
+    """
+    for clip_annotation in dataset:
+        for sound_event_annotation in clip_annotation.sound_events:
+            if apply_filter:
+                if not targets.filter(sound_event_annotation):
+                    continue
+
+            if apply_transform:
+                sound_event_annotation = targets.transform(
+                    sound_event_annotation
+                )
+
+            class_name = targets.encode(sound_event_annotation)
+            if class_name is None and exclude_generic:
+                continue
+
+            yield class_name, sound_event_annotation

batdetect2/postprocess/remapping.py

@@ -84,7 +84,7 @@ def features_to_xarray(
     freqs = np.linspace(min_freq, max_freq, height, endpoint=False)
 
     return xr.DataArray(
-        data=features.detach().numpy(),
+        data=features.detach().cpu().numpy(),
         dims=[
             Dimensions.feature.value,
             Dimensions.frequency.value,
@@ -157,7 +157,7 @@ def detection_to_xarray(
     freqs = np.linspace(min_freq, max_freq, height, endpoint=False)
 
     return xr.DataArray(
-        data=detection.squeeze(dim=0).detach().numpy(),
+        data=detection.squeeze(dim=0).detach().cpu().numpy(),
         dims=[
             Dimensions.frequency.value,
             Dimensions.time.value,
@@ -233,7 +233,7 @@ def classification_to_xarray(
     freqs = np.linspace(min_freq, max_freq, height, endpoint=False)
 
     return xr.DataArray(
-        data=classes.detach().numpy(),
+        data=classes.detach().cpu().numpy(),
         dims=[
             "category",
             Dimensions.frequency.value,
@@ -302,7 +302,7 @@ def sizes_to_xarray(
     freqs = np.linspace(min_freq, max_freq, height, endpoint=False)
 
     return xr.DataArray(
-        data=sizes.detach().numpy(),
+        data=sizes.detach().cpu().numpy(),
         dims=[
             "dimension",
             Dimensions.frequency.value,

batdetect2/preprocess/spectrogram.py

@@ -19,7 +19,7 @@ reproducible spectrogram generation consistent between training and inference.
 The core computation is performed by `compute_spectrogram`.
 """
 
-from typing import Literal, Optional, Union
+from typing import Callable, Literal, Optional, Union
 
 import numpy as np
 import xarray as xr
@@ -327,44 +327,45 @@ def compute_spectrogram(
     """
     config = config or SpectrogramConfig()
 
-    spec = stft(
+    with xr.set_options(keep_attrs=True):
-        wav,
+        spec = stft(
-        window_duration=config.stft.window_duration,
+            wav,
-        window_overlap=config.stft.window_overlap,
+            window_duration=config.stft.window_duration,
-        window_fn=config.stft.window_fn,
+            window_overlap=config.stft.window_overlap,
-    )
+            window_fn=config.stft.window_fn,
-
-    spec = crop_spectrogram_frequencies(
-        spec,
-        min_freq=config.frequencies.min_freq,
-        max_freq=config.frequencies.max_freq,
-    )
-
-    if config.pcen:
-        spec = apply_pcen(
-            spec,
-            time_constant=config.pcen.time_constant,
-            gain=config.pcen.gain,
-            power=config.pcen.power,
-            bias=config.pcen.bias,
         )
 
-    spec = scale_spectrogram(spec, scale=config.scale)
+        spec = crop_spectrogram_frequencies(
-
-    if config.spectral_mean_substraction:
-        spec = remove_spectral_mean(spec)
-
-    if config.size:
-        spec = resize_spectrogram(
             spec,
-            height=config.size.height,
+            min_freq=config.frequencies.min_freq,
-            resize_factor=config.size.resize_factor,
+            max_freq=config.frequencies.max_freq,
         )
 
-    if config.peak_normalize:
+        if config.pcen:
-        spec = ops.scale(spec, 1 / (10e-6 + np.max(spec)))
+            spec = apply_pcen(
+                spec,
+                time_constant=config.pcen.time_constant,
+                gain=config.pcen.gain,
+                power=config.pcen.power,
+                bias=config.pcen.bias,
+            )
 
-    return spec.astype(dtype)
+        spec = scale_spectrogram(spec, scale=config.scale)
+
+        if config.spectral_mean_substraction:
+            spec = remove_spectral_mean(spec)
+
+        if config.size:
+            spec = resize_spectrogram(
+                spec,
+                height=config.size.height,
+                resize_factor=config.size.resize_factor,
+            )
+
+        if config.peak_normalize:
+            spec = ops.scale(spec, 1 / (10e-6 + np.max(spec)))
+
+        return spec.astype(dtype)
 
 
 def crop_spectrogram_frequencies(
@@ -477,7 +478,7 @@ def scale_spectrogram(
     """Apply final amplitude scaling/representation to the spectrogram.
 
     Converts the input magnitude spectrogram based on the `scale` type:
-    - "dB": Applies logarithmic scaling `log1p(C * S)`.
+    - "dB": Applies logarithmic scaling `log10(S)`.
     - "power": Squares the magnitude values `S^2`.
     - "amplitude": Returns the input magnitude values `S` unchanged.
 
@@ -496,7 +497,7 @@ def scale_spectrogram(
         Spectrogram with the specified amplitude scaling applied.
     """
     if scale == "dB":
-        return scale_log(spec, dtype=dtype)
+        return arrays.to_db(spec).astype(dtype)
 
     if scale == "power":
         return spec**2
@@ -561,12 +562,13 @@ def apply_pcen(
 def scale_log(
     spec: xr.DataArray,
     dtype: DTypeLike = np.float32,  # type: ignore
+    ref: Union[float, Callable] = np.max,
+    amin: float = 1e-10,
+    top_db: Optional[float] = 80.0,
 ) -> xr.DataArray:
     """Apply logarithmic scaling to a magnitude spectrogram.
 
-    Calculates `log(1 + C * S)`, where S is the input magnitude spectrogram
+    Calculates `log10(S)`, where S is the input magnitude spectrogram.
-    and C is a scaling factor derived from the original STFT parameters
-    (sample rate, N-FFT, window function) stored in `spec.attrs`.
 
     Parameters
     ----------
@@ -587,12 +589,29 @@ def scale_log(
         If required attributes are missing from `spec.attrs`.
     ValueError
         If attributes are non-numeric or window function is invalid.
+
+
+    Notes
+    -----
+    Implementation mainly taken from librosa `power_to_db` function
     """
-    samplerate = spec.attrs["original_samplerate"]
+
-    nfft = spec.attrs["nfft"]
+    if callable(ref):
-    log_scaling = 2 / (samplerate * (np.abs(np.hanning(nfft)) ** 2).sum())
+        ref_value = ref(spec)
+    else:
+        ref_value = np.abs(ref)
+
+    log_spec = 10.0 * np.log10(np.maximum(amin, spec)) - np.log10(
+        np.maximum(amin, ref_value)
+    )
+
+    if top_db is not None:
+        if top_db < 0:
+            raise ValueError("top_db must be non-negative")
+        log_spec = np.maximum(log_spec, log_spec.max() - top_db)
+
     return xr.DataArray(
-        data=np.log1p(log_scaling * spec).astype(dtype),
+        data=log_spec.astype(dtype),
         dims=spec.dims,
         coords=spec.coords,
         attrs=spec.attrs,

batdetect2/train/dataset.py

@@ -9,6 +9,7 @@ from torch.utils.data import Dataset
 
 from batdetect2.train.augmentations import Augmentation
 from batdetect2.train.types import ClipperProtocol, TrainExample
+from batdetect2.utils.tensors import adjust_width
 
 __all__ = [
     "LabeledDataset",
@@ -87,6 +88,26 @@ class LabeledDataset(Dataset):
         )
 
 
+def collate_fn(batch: List[TrainExample]):
+    width = 512
+
+    return TrainExample(
+        spec=torch.stack([adjust_width(x.spec, width) for x in batch]),
+        detection_heatmap=torch.stack(
+            [adjust_width(x.detection_heatmap, width) for x in batch]
+        ),
+        class_heatmap=torch.stack(
+            [adjust_width(x.class_heatmap, width) for x in batch]
+        ),
+        size_heatmap=torch.stack(
+            [adjust_width(x.size_heatmap, width) for x in batch]
+        ),
+        idx=torch.stack([x.idx for x in batch]),
+        start_time=torch.stack([x.start_time for x in batch]),
+        end_time=torch.stack([x.end_time for x in batch]),
+    )
+
+
 def list_preprocessed_files(
     directory: data.PathLike, extension: str = ".nc"
 ) -> List[Path]:

batdetect2/train/preprocess.py

@@ -276,8 +276,11 @@ def preprocess_single_annotation(
             labeller=labeller,
         )
     except Exception as error:
-        raise RuntimeError(
+        logger.error(
-            f"Failed to process annotation: {clip_annotation.uuid}"
+            "Failed to process annotation: {uuid}. Error {error}",
-        ) from error
+            uuid=clip_annotation.uuid,
+            error=error,
+        )
+        return
 
     _save_xr_dataset_to_file(sample, path)

batdetect2/train/train.py

@@ -17,7 +17,11 @@ from batdetect2.train.augmentations import (
 )
 from batdetect2.train.clips import build_clipper
 from batdetect2.train.config import TrainingConfig
-from batdetect2.train.dataset import LabeledDataset, RandomExampleSource
+from batdetect2.train.dataset import (
+    LabeledDataset,
+    RandomExampleSource,
+    collate_fn,
+)
 from batdetect2.train.lightning import TrainingModule
 from batdetect2.train.losses import build_loss
 
@@ -88,6 +92,7 @@ def train(
         batch_size=config.batch_size,
         shuffle=True,
         num_workers=train_workers,
+        collate_fn=collate_fn,
     )
 
     val_dataloader = None
@@ -101,6 +106,7 @@ def train(
             batch_size=config.batch_size,
             shuffle=False,
             num_workers=val_workers,
+            collate_fn=collate_fn,
         )
 
     trainer.fit(

requirements-dev.lock

@@ -1,540 +0,0 @@
-# generated by rye
-# use `rye lock` or `rye sync` to update this lockfile
-#
-# last locked with the following flags:
-#   pre: false
-#   features: []
-#   all-features: false
-#   with-sources: false
-#   generate-hashes: false
-#   universal: false
-
--e file:.
-absl-py==2.1.0
-    # via tensorboard
-affine==2.4.0
-    # via rasterio
-aiobotocore==2.12.3
-    # via s3fs
-aiohttp==3.9.5
-    # via aiobotocore
-    # via fsspec
-    # via lightning
-    # via s3fs
-aioitertools==0.11.0
-    # via aiobotocore
-aiosignal==1.3.1
-    # via aiohttp
-annotated-types==0.6.0
-    # via pydantic
-antlr4-python3-runtime==4.9.3
-    # via hydra-core
-    # via omegaconf
-anyio==4.3.0
-    # via starlette
-arrow==1.3.0
-    # via lightning
-asttokens==2.4.1
-    # via stack-data
-async-timeout==4.0.3
-    # via aiohttp
-    # via redis
-attrs==23.2.0
-    # via aiohttp
-    # via rasterio
-audioread==3.0.1
-    # via librosa
-backcall==0.2.0
-    # via ipython
-backoff==2.2.1
-    # via lightning
-beautifulsoup4==4.12.3
-    # via lightning
-bitsandbytes==0.41.0
-    # via lightning
-blessed==1.20.0
-    # via inquirer
-boto3==1.34.69
-    # via lightning-cloud
-botocore==1.34.69
-    # via aiobotocore
-    # via boto3
-    # via s3transfer
-certifi==2024.2.2
-    # via netcdf4
-    # via rasterio
-    # via requests
-cf-xarray==0.9.0
-    # via batdetect2
-cffi==1.16.0
-    # via soundfile
-cftime==1.6.3
-    # via netcdf4
-charset-normalizer==3.3.2
-    # via requests
-click==8.1.7
-    # via batdetect2
-    # via click-plugins
-    # via cligj
-    # via lightning
-    # via lightning-cloud
-    # via rasterio
-    # via uvicorn
-click-plugins==1.1.1
-    # via rasterio
-cligj==0.7.2
-    # via rasterio
-comm==0.2.2
-    # via ipykernel
-contourpy==1.1.1
-    # via matplotlib
-croniter==1.4.1
-    # via lightning
-cycler==0.12.1
-    # via matplotlib
-cython==3.0.10
-    # via soundevent
-dateutils==0.6.12
-    # via lightning
-debugpy==1.8.1
-    # via ipykernel
-decorator==5.1.1
-    # via ipython
-    # via librosa
-deepdiff==6.7.1
-    # via lightning
-dnspython==2.6.1
-    # via email-validator
-docker==6.1.3
-    # via lightning
-docstring-parser==0.16
-    # via jsonargparse
-editor==1.6.6
-    # via inquirer
-email-validator==2.1.1
-    # via soundevent
-exceptiongroup==1.2.1
-    # via anyio
-    # via pytest
-executing==2.0.1
-    # via stack-data
-fastapi==0.110.2
-    # via lightning
-    # via lightning-cloud
-filelock==3.13.4
-    # via torch
-    # via triton
-fonttools==4.51.0
-    # via matplotlib
-frozenlist==1.4.1
-    # via aiohttp
-    # via aiosignal
-fsspec==2023.12.2
-    # via lightning
-    # via lightning-fabric
-    # via pytorch-lightning
-    # via s3fs
-    # via torch
-grpcio==1.62.2
-    # via tensorboard
-h11==0.14.0
-    # via uvicorn
-hydra-core==1.3.2
-    # via lightning
-idna==3.7
-    # via anyio
-    # via email-validator
-    # via requests
-    # via yarl
-importlib-metadata==7.1.0
-    # via jupyter-client
-    # via markdown
-    # via rasterio
-importlib-resources==6.4.0
-    # via matplotlib
-    # via typeshed-client
-iniconfig==2.0.0
-    # via pytest
-inquirer==3.2.4
-    # via lightning
-ipykernel==6.29.4
-ipython==8.12.3
-    # via ipykernel
-jedi==0.19.1
-    # via ipython
-jinja2==3.1.3
-    # via lightning
-    # via torch
-jmespath==1.0.1
-    # via boto3
-    # via botocore
-joblib==1.4.0
-    # via librosa
-    # via scikit-learn
-jsonargparse==4.28.0
-    # via lightning
-jupyter-client==8.6.1
-    # via ipykernel
-jupyter-core==5.7.2
-    # via ipykernel
-    # via jupyter-client
-kiwisolver==1.4.5
-    # via matplotlib
-lazy-loader==0.4
-    # via librosa
-librosa==0.10.1
-    # via batdetect2
-lightning==2.2.2
-    # via batdetect2
-lightning-api-access==0.0.5
-    # via lightning
-lightning-cloud==0.5.65
-    # via lightning
-lightning-fabric==2.2.2
-    # via lightning
-lightning-utilities==0.11.2
-    # via lightning
-    # via lightning-fabric
-    # via pytorch-lightning
-    # via torchmetrics
-llvmlite==0.41.1
-    # via numba
-markdown==3.6
-    # via tensorboard
-markdown-it-py==3.0.0
-    # via rich
-markupsafe==2.1.5
-    # via jinja2
-    # via werkzeug
-matplotlib==3.7.5
-    # via batdetect2
-    # via lightning
-    # via soundevent
-matplotlib-inline==0.1.7
-    # via ipykernel
-    # via ipython
-mdurl==0.1.2
-    # via markdown-it-py
-mpmath==1.3.0
-    # via sympy
-msgpack==1.0.8
-    # via librosa
-multidict==6.0.5
-    # via aiohttp
-    # via yarl
-nest-asyncio==1.6.0
-    # via ipykernel
-netcdf4==1.6.5
-    # via batdetect2
-networkx==3.1
-    # via torch
-numba==0.58.1
-    # via librosa
-numpy==1.24.4
-    # via batdetect2
-    # via cftime
-    # via contourpy
-    # via librosa
-    # via lightning
-    # via lightning-fabric
-    # via matplotlib
-    # via netcdf4
-    # via numba
-    # via onnx
-    # via pandas
-    # via pytorch-lightning
-    # via rasterio
-    # via scikit-learn
-    # via scipy
-    # via shapely
-    # via snuggs
-    # via soxr
-    # via tensorboard
-    # via tensorboardx
-    # via torchmetrics
-    # via torchvision
-    # via xarray
-nvidia-cublas-cu12==12.1.3.1
-    # via nvidia-cudnn-cu12
-    # via nvidia-cusolver-cu12
-    # via torch
-nvidia-cuda-cupti-cu12==12.1.105
-    # via torch
-nvidia-cuda-nvrtc-cu12==12.1.105
-    # via torch
-nvidia-cuda-runtime-cu12==12.1.105
-    # via torch
-nvidia-cudnn-cu12==8.9.2.26
-    # via torch
-nvidia-cufft-cu12==11.0.2.54
-    # via torch
-nvidia-curand-cu12==10.3.2.106
-    # via torch
-nvidia-cusolver-cu12==11.4.5.107
-    # via torch
-nvidia-cusparse-cu12==12.1.0.106
-    # via nvidia-cusolver-cu12
-    # via torch
-nvidia-nccl-cu12==2.19.3
-    # via torch
-nvidia-nvjitlink-cu12==12.4.127
-    # via nvidia-cusolver-cu12
-    # via nvidia-cusparse-cu12
-nvidia-nvtx-cu12==12.1.105
-    # via torch
-omegaconf==2.3.0
-    # via hydra-core
-    # via lightning
-onnx==1.16.0
-    # via batdetect2
-ordered-set==4.1.0
-    # via deepdiff
-packaging==24.0
-    # via docker
-    # via hydra-core
-    # via ipykernel
-    # via lazy-loader
-    # via lightning
-    # via lightning-fabric
-    # via lightning-utilities
-    # via matplotlib
-    # via pooch
-    # via pytest
-    # via pytorch-lightning
-    # via tensorboardx
-    # via torchmetrics
-    # via xarray
-pandas==2.0.3
-    # via batdetect2
-    # via xarray
-parso==0.8.4
-    # via jedi
-pexpect==4.9.0
-    # via ipython
-pickleshare==0.7.5
-    # via ipython
-pillow==10.3.0
-    # via matplotlib
-    # via torchvision
-platformdirs==4.2.0
-    # via jupyter-core
-    # via pooch
-pluggy==1.4.0
-    # via pytest
-pooch==1.8.1
-    # via librosa
-prompt-toolkit==3.0.43
-    # via ipython
-protobuf==5.26.1
-    # via onnx
-    # via tensorboard
-    # via tensorboardx
-psutil==5.9.8
-    # via ipykernel
-    # via lightning
-ptyprocess==0.7.0
-    # via pexpect
-pure-eval==0.2.2
-    # via stack-data
-pycparser==2.22
-    # via cffi
-pydantic==2.7.0
-    # via fastapi
-    # via lightning
-    # via soundevent
-pydantic-core==2.18.1
-    # via pydantic
-pygments==2.17.2
-    # via ipython
-    # via rich
-pyjwt==2.8.0
-    # via lightning-cloud
-pyparsing==3.1.2
-    # via matplotlib
-    # via snuggs
-pytest==8.1.1
-python-dateutil==2.9.0.post0
-    # via arrow
-    # via botocore
-    # via croniter
-    # via dateutils
-    # via jupyter-client
-    # via matplotlib
-    # via pandas
-python-multipart==0.0.9
-    # via lightning
-    # via lightning-cloud
-pytorch-lightning==2.2.2
-    # via batdetect2
-    # via lightning
-pytz==2024.1
-    # via dateutils
-    # via pandas
-pyyaml==6.0.1
-    # via jsonargparse
-    # via lightning
-    # via omegaconf
-    # via pytorch-lightning
-pyzmq==26.0.0
-    # via ipykernel
-    # via jupyter-client
-rasterio==1.3.10
-    # via soundevent
-readchar==4.0.6
-    # via inquirer
-redis==5.0.4
-    # via lightning
-requests==2.31.0
-    # via docker
-    # via fsspec
-    # via lightning
-    # via lightning-cloud
-    # via pooch
-rich==13.7.1
-    # via lightning
-    # via lightning-cloud
-runs==1.2.2
-    # via editor
-s3fs==2023.12.2
-    # via lightning
-s3transfer==0.10.1
-    # via boto3
-scikit-learn==1.3.2
-    # via batdetect2
-    # via librosa
-scipy==1.10.1
-    # via batdetect2
-    # via librosa
-    # via scikit-learn
-    # via soundevent
-setuptools==69.5.1
-    # via lightning-utilities
-    # via rasterio
-    # via readchar
-    # via tensorboard
-shapely==2.0.3
-    # via soundevent
-six==1.16.0
-    # via asttokens
-    # via blessed
-    # via lightning-cloud
-    # via python-dateutil
-    # via tensorboard
-sniffio==1.3.1
-    # via anyio
-snuggs==1.4.7
-    # via rasterio
-soundevent==2.0.1
-    # via batdetect2
-soundfile==0.12.1
-    # via librosa
-    # via soundevent
-soupsieve==2.5
-    # via beautifulsoup4
-soxr==0.3.7
-    # via librosa
-stack-data==0.6.3
-    # via ipython
-starlette==0.37.2
-    # via fastapi
-    # via lightning
-sympy==1.12
-    # via torch
-tensorboard==2.16.2
-    # via batdetect2
-tensorboard-data-server==0.7.2
-    # via tensorboard
-tensorboardx==2.6.2.2
-    # via lightning
-threadpoolctl==3.4.0
-    # via scikit-learn
-tomli==2.0.1
-    # via pytest
-torch==2.2.2
-    # via batdetect2
-    # via lightning
-    # via lightning-fabric
-    # via pytorch-lightning
-    # via torchaudio
-    # via torchmetrics
-    # via torchvision
-torchaudio==2.2.2
-    # via batdetect2
-torchmetrics==1.3.2
-    # via lightning
-    # via pytorch-lightning
-torchvision==0.17.2
-    # via batdetect2
-tornado==6.4
-    # via ipykernel
-    # via jupyter-client
-tqdm==4.66.2
-    # via batdetect2
-    # via lightning
-    # via pytorch-lightning
-traitlets==5.14.2
-    # via comm
-    # via ipykernel
-    # via ipython
-    # via jupyter-client
-    # via jupyter-core
-    # via lightning
-    # via matplotlib-inline
-triton==2.2.0
-    # via torch
-types-python-dateutil==2.9.0.20240316
-    # via arrow
-typeshed-client==2.5.1
-    # via jsonargparse
-typing-extensions==4.11.0
-    # via aioitertools
-    # via anyio
-    # via fastapi
-    # via ipython
-    # via jsonargparse
-    # via librosa
-    # via lightning
-    # via lightning-fabric
-    # via lightning-utilities
-    # via pydantic
-    # via pydantic-core
-    # via pytorch-lightning
-    # via starlette
-    # via torch
-    # via typeshed-client
-    # via uvicorn
-tzdata==2024.1
-    # via pandas
-urllib3==1.26.18
-    # via botocore
-    # via docker
-    # via lightning
-    # via lightning-cloud
-    # via requests
-uvicorn==0.29.0
-    # via lightning
-    # via lightning-cloud
-wcwidth==0.2.13
-    # via blessed
-    # via prompt-toolkit
-websocket-client==1.7.0
-    # via docker
-    # via lightning
-    # via lightning-cloud
-websockets==11.0.3
-    # via lightning
-werkzeug==3.0.2
-    # via tensorboard
-wrapt==1.16.0
-    # via aiobotocore
-xarray==2023.1.0
-    # via cf-xarray
-    # via soundevent
-xmod==1.8.1
-    # via editor
-    # via runs
-yarl==1.9.4
-    # via aiohttp
-zipp==3.18.1
-    # via importlib-metadata
-    # via importlib-resources

requirements.lock

@@ -1,470 +0,0 @@
-# generated by rye
-# use `rye lock` or `rye sync` to update this lockfile
-#
-# last locked with the following flags:
-#   pre: false
-#   features: []
-#   all-features: false
-#   with-sources: false
-#   generate-hashes: false
-#   universal: false
-
--e file:.
-absl-py==2.1.0
-    # via tensorboard
-affine==2.4.0
-    # via rasterio
-aiobotocore==2.12.3
-    # via s3fs
-aiohttp==3.9.5
-    # via aiobotocore
-    # via fsspec
-    # via lightning
-    # via s3fs
-aioitertools==0.11.0
-    # via aiobotocore
-aiosignal==1.3.1
-    # via aiohttp
-annotated-types==0.6.0
-    # via pydantic
-antlr4-python3-runtime==4.9.3
-    # via hydra-core
-    # via omegaconf
-anyio==4.3.0
-    # via starlette
-arrow==1.3.0
-    # via lightning
-async-timeout==4.0.3
-    # via aiohttp
-    # via redis
-attrs==23.2.0
-    # via aiohttp
-    # via rasterio
-audioread==3.0.1
-    # via librosa
-backoff==2.2.1
-    # via lightning
-beautifulsoup4==4.12.3
-    # via lightning
-bitsandbytes==0.41.0
-    # via lightning
-blessed==1.20.0
-    # via inquirer
-boto3==1.34.69
-    # via lightning-cloud
-botocore==1.34.69
-    # via aiobotocore
-    # via boto3
-    # via s3transfer
-certifi==2024.2.2
-    # via netcdf4
-    # via rasterio
-    # via requests
-cf-xarray==0.9.0
-    # via batdetect2
-cffi==1.16.0
-    # via soundfile
-cftime==1.6.3
-    # via netcdf4
-charset-normalizer==3.3.2
-    # via requests
-click==8.1.7
-    # via batdetect2
-    # via click-plugins
-    # via cligj
-    # via lightning
-    # via lightning-cloud
-    # via rasterio
-    # via uvicorn
-click-plugins==1.1.1
-    # via rasterio
-cligj==0.7.2
-    # via rasterio
-contourpy==1.1.1
-    # via matplotlib
-croniter==1.4.1
-    # via lightning
-cycler==0.12.1
-    # via matplotlib
-cython==3.0.10
-    # via soundevent
-dateutils==0.6.12
-    # via lightning
-decorator==5.1.1
-    # via librosa
-deepdiff==6.7.1
-    # via lightning
-dnspython==2.6.1
-    # via email-validator
-docker==6.1.3
-    # via lightning
-docstring-parser==0.16
-    # via jsonargparse
-editor==1.6.6
-    # via inquirer
-email-validator==2.1.1
-    # via soundevent
-exceptiongroup==1.2.1
-    # via anyio
-fastapi==0.110.2
-    # via lightning
-    # via lightning-cloud
-filelock==3.13.4
-    # via torch
-    # via triton
-fonttools==4.51.0
-    # via matplotlib
-frozenlist==1.4.1
-    # via aiohttp
-    # via aiosignal
-fsspec==2023.12.2
-    # via lightning
-    # via lightning-fabric
-    # via pytorch-lightning
-    # via s3fs
-    # via torch
-grpcio==1.62.2
-    # via tensorboard
-h11==0.14.0
-    # via uvicorn
-hydra-core==1.3.2
-    # via lightning
-idna==3.7
-    # via anyio
-    # via email-validator
-    # via requests
-    # via yarl
-importlib-metadata==7.1.0
-    # via markdown
-    # via rasterio
-importlib-resources==6.4.0
-    # via matplotlib
-    # via typeshed-client
-inquirer==3.2.4
-    # via lightning
-jinja2==3.1.3
-    # via lightning
-    # via torch
-jmespath==1.0.1
-    # via boto3
-    # via botocore
-joblib==1.4.0
-    # via librosa
-    # via scikit-learn
-jsonargparse==4.28.0
-    # via lightning
-kiwisolver==1.4.5
-    # via matplotlib
-lazy-loader==0.4
-    # via librosa
-librosa==0.10.1
-    # via batdetect2
-lightning==2.2.2
-    # via batdetect2
-lightning-api-access==0.0.5
-    # via lightning
-lightning-cloud==0.5.65
-    # via lightning
-lightning-fabric==2.2.2
-    # via lightning
-lightning-utilities==0.11.2
-    # via lightning
-    # via lightning-fabric
-    # via pytorch-lightning
-    # via torchmetrics
-llvmlite==0.41.1
-    # via numba
-markdown==3.6
-    # via tensorboard
-markdown-it-py==3.0.0
-    # via rich
-markupsafe==2.1.5
-    # via jinja2
-    # via werkzeug
-matplotlib==3.7.5
-    # via batdetect2
-    # via lightning
-    # via soundevent
-mdurl==0.1.2
-    # via markdown-it-py
-mpmath==1.3.0
-    # via sympy
-msgpack==1.0.8
-    # via librosa
-multidict==6.0.5
-    # via aiohttp
-    # via yarl
-netcdf4==1.6.5
-    # via batdetect2
-networkx==3.1
-    # via torch
-numba==0.58.1
-    # via librosa
-numpy==1.24.4
-    # via batdetect2
-    # via cftime
-    # via contourpy
-    # via librosa
-    # via lightning
-    # via lightning-fabric
-    # via matplotlib
-    # via netcdf4
-    # via numba
-    # via onnx
-    # via pandas
-    # via pytorch-lightning
-    # via rasterio
-    # via scikit-learn
-    # via scipy
-    # via shapely
-    # via snuggs
-    # via soxr
-    # via tensorboard
-    # via tensorboardx
-    # via torchmetrics
-    # via torchvision
-    # via xarray
-nvidia-cublas-cu12==12.1.3.1
-    # via nvidia-cudnn-cu12
-    # via nvidia-cusolver-cu12
-    # via torch
-nvidia-cuda-cupti-cu12==12.1.105
-    # via torch
-nvidia-cuda-nvrtc-cu12==12.1.105
-    # via torch
-nvidia-cuda-runtime-cu12==12.1.105
-    # via torch
-nvidia-cudnn-cu12==8.9.2.26
-    # via torch
-nvidia-cufft-cu12==11.0.2.54
-    # via torch
-nvidia-curand-cu12==10.3.2.106
-    # via torch
-nvidia-cusolver-cu12==11.4.5.107
-    # via torch
-nvidia-cusparse-cu12==12.1.0.106
-    # via nvidia-cusolver-cu12
-    # via torch
-nvidia-nccl-cu12==2.19.3
-    # via torch
-nvidia-nvjitlink-cu12==12.4.127
-    # via nvidia-cusolver-cu12
-    # via nvidia-cusparse-cu12
-nvidia-nvtx-cu12==12.1.105
-    # via torch
-omegaconf==2.3.0
-    # via hydra-core
-    # via lightning
-onnx==1.16.0
-    # via batdetect2
-ordered-set==4.1.0
-    # via deepdiff
-packaging==24.0
-    # via docker
-    # via hydra-core
-    # via lazy-loader
-    # via lightning
-    # via lightning-fabric
-    # via lightning-utilities
-    # via matplotlib
-    # via pooch
-    # via pytorch-lightning
-    # via tensorboardx
-    # via torchmetrics
-    # via xarray
-pandas==2.0.3
-    # via batdetect2
-    # via xarray
-pillow==10.3.0
-    # via matplotlib
-    # via torchvision
-platformdirs==4.2.0
-    # via pooch
-pooch==1.8.1
-    # via librosa
-protobuf==5.26.1
-    # via onnx
-    # via tensorboard
-    # via tensorboardx
-psutil==5.9.8
-    # via lightning
-pycparser==2.22
-    # via cffi
-pydantic==2.7.0
-    # via fastapi
-    # via lightning
-    # via soundevent
-pydantic-core==2.18.1
-    # via pydantic
-pygments==2.17.2
-    # via rich
-pyjwt==2.8.0
-    # via lightning-cloud
-pyparsing==3.1.2
-    # via matplotlib
-    # via snuggs
-python-dateutil==2.9.0.post0
-    # via arrow
-    # via botocore
-    # via croniter
-    # via dateutils
-    # via matplotlib
-    # via pandas
-python-multipart==0.0.9
-    # via lightning
-    # via lightning-cloud
-pytorch-lightning==2.2.2
-    # via batdetect2
-    # via lightning
-pytz==2024.1
-    # via dateutils
-    # via pandas
-pyyaml==6.0.1
-    # via jsonargparse
-    # via lightning
-    # via omegaconf
-    # via pytorch-lightning
-rasterio==1.3.10
-    # via soundevent
-readchar==4.0.6
-    # via inquirer
-redis==5.0.4
-    # via lightning
-requests==2.31.0
-    # via docker
-    # via fsspec
-    # via lightning
-    # via lightning-cloud
-    # via pooch
-rich==13.7.1
-    # via lightning
-    # via lightning-cloud
-runs==1.2.2
-    # via editor
-s3fs==2023.12.2
-    # via lightning
-s3transfer==0.10.1
-    # via boto3
-scikit-learn==1.3.2
-    # via batdetect2
-    # via librosa
-scipy==1.10.1
-    # via batdetect2
-    # via librosa
-    # via scikit-learn
-    # via soundevent
-setuptools==69.5.1
-    # via lightning-utilities
-    # via rasterio
-    # via readchar
-    # via tensorboard
-shapely==2.0.3
-    # via soundevent
-six==1.16.0
-    # via blessed
-    # via lightning-cloud
-    # via python-dateutil
-    # via tensorboard
-sniffio==1.3.1
-    # via anyio
-snuggs==1.4.7
-    # via rasterio
-soundevent==2.0.1
-    # via batdetect2
-soundfile==0.12.1
-    # via librosa
-    # via soundevent
-soupsieve==2.5
-    # via beautifulsoup4
-soxr==0.3.7
-    # via librosa
-starlette==0.37.2
-    # via fastapi
-    # via lightning
-sympy==1.12
-    # via torch
-tensorboard==2.16.2
-    # via batdetect2
-tensorboard-data-server==0.7.2
-    # via tensorboard
-tensorboardx==2.6.2.2
-    # via lightning
-threadpoolctl==3.4.0
-    # via scikit-learn
-torch==2.2.2
-    # via batdetect2
-    # via lightning
-    # via lightning-fabric
-    # via pytorch-lightning
-    # via torchaudio
-    # via torchmetrics
-    # via torchvision
-torchaudio==2.2.2
-    # via batdetect2
-torchmetrics==1.3.2
-    # via lightning
-    # via pytorch-lightning
-torchvision==0.17.2
-    # via batdetect2
-tqdm==4.66.2
-    # via batdetect2
-    # via lightning
-    # via pytorch-lightning
-traitlets==5.14.3
-    # via lightning
-triton==2.2.0
-    # via torch
-types-python-dateutil==2.9.0.20240316
-    # via arrow
-typeshed-client==2.5.1
-    # via jsonargparse
-typing-extensions==4.11.0
-    # via aioitertools
-    # via anyio
-    # via fastapi
-    # via jsonargparse
-    # via librosa
-    # via lightning
-    # via lightning-fabric
-    # via lightning-utilities
-    # via pydantic
-    # via pydantic-core
-    # via pytorch-lightning
-    # via starlette
-    # via torch
-    # via typeshed-client
-    # via uvicorn
-tzdata==2024.1
-    # via pandas
-urllib3==1.26.18
-    # via botocore
-    # via docker
-    # via lightning
-    # via lightning-cloud
-    # via requests
-uvicorn==0.29.0
-    # via lightning
-    # via lightning-cloud
-wcwidth==0.2.13
-    # via blessed
-websocket-client==1.7.0
-    # via docker
-    # via lightning
-    # via lightning-cloud
-websockets==11.0.3
-    # via lightning
-werkzeug==3.0.2
-    # via tensorboard
-wrapt==1.16.0
-    # via aiobotocore
-xarray==2023.1.0
-    # via cf-xarray
-    # via soundevent
-xmod==1.8.1
-    # via editor
-    # via runs
-yarl==1.9.4
-    # via aiohttp
-zipp==3.18.1
-    # via importlib-metadata
-    # via importlib-resources