Upload folder using huggingface_hub

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/README.md

@@ -0,0 +1,5 @@
+system=default_seed-0
+data=birdclef2025_15s
+train=birdclef2025_convnext_tiny
+model=birdclef2025_convnext_tiny
+criterion=birdclef2025_categorical_cross_entropy

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/log/20250429-091735/.hydra/config.yaml

@@ -0,0 +1,221 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: null
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: null
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: ${..audio.sample_rate}
+    hop_length: 1253
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: ${..train.audio_key}
+      sample_rate_key: ${..train.sample_rate_key}
+      label_name_key: ${..train.label_name_key}
+      filename_key: ${..train.filename_key}
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform: ${data.melspectrogram}
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: ${data.audio.sample_rate}
+          duration: ${data.audio.duration}
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+        melspectrogram_key: ${.composer.melspectrogram_key}
+        label_index_key: ${.composer.label_index_key}
+        alpha: 0.4
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: ${....train.collate_fn.composer._target_}
+          melspectrogram_transform: ${....train.collate_fn.composer.melspectrogram_transform}
+          audio_key: ${....train.collate_fn.composer.audio_key}
+          sample_rate_key: ${....train.collate_fn.composer.sample_rate_key}
+          label_name_key: ${....train.collate_fn.composer.label_name_key}
+          filename_key: ${....train.collate_fn.composer.filename_key}
+          waveform_key: ${....train.collate_fn.composer.waveform_key}
+          melspectrogram_key: ${....train.collate_fn.composer.melspectrogram_key}
+          label_index_key: ${....train.collate_fn.composer.label_index_key}
+          sample_rate: ${....train.collate_fn.composer.sample_rate}
+          duration: ${....train.collate_fn.composer.duration}
+          decode_audio_as_waveform: ${....train.collate_fn.composer.decode_audio_as_waveform}
+          decode_audio_as_monoral: ${....train.collate_fn.composer.decode_audio_as_monoral}
+          training: false
+        melspectrogram_key: ${...train.collate_fn.composer.melspectrogram_key}
+        label_index_key: ${...train.collate_fn.composer.label_index_key}
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: ${....dataloader.train.collate_fn.composer.melspectrogram_key}
+      output: logit
+    validation: ${.train}
+    inference: ${.validation}
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250429-091731
+    tensorboard_dir: ./tensorboard/20250429-091731
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ${...exp_dir}/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ${...exp_dir}/model/epoch{epoch}.pth
+      last:
+        path: ${...exp_dir}/model/last.pth
+      best_epoch:
+        path: ${...exp_dir}/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ${.exp_dir}/inference
+    audio:
+      sample_rate: ${data.audio.sample_rate}
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+model:
+  _target_: birdclef2025.models.ConvNeXtTiny
+  weights: ${const:torchvision.models.ConvNeXt_Tiny_Weights.IMAGENET1K_V1}
+  num_classes: ${const:birdclef2025.utils.data.birdclef.num_birdclef2025_primary_labels}
+optimizer:
+  _target_: torch.optim.AdamW
+  lr: 0.0001
+  weight_decay: 0.05
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: ${train.dataloader.train.collate_fn.composer.label_index_key}

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/log/20250429-091735/.hydra/hydra.yaml

@@ -0,0 +1,189 @@
+hydra:
+  run:
+    dir: ./exp/20250429-091731/log/20250429-091735
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=./exp/20250429-091731/log/20250429-091735
+    - hydra.mode=RUN
+    task:
+    - system=cuda
+    - preprocess=birdclef2025
+    - data=birdclef2025_15s
+    - train=birdclef2025_convnext_tiny
+    - model=birdclef2025_convnext_tiny
+    - optimizer=adamw_1e-4_decay_5e-2
+    - lr_scheduler=none
+    - criterion=birdclef2025_categorical_cross_entropy
+    - preprocess.dump_format=birdclef2025
+    - train.dataset.train.list_path=dump/birdclef2025_15s/list/train.txt
+    - train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+    - train.dataset.validation.list_path=dump/birdclef2025_15s/list/validation.txt
+    - train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+    - train.resume.continue_from=
+    - train.output.exp_dir=./exp/20250429-091731
+    - train.output.tensorboard_dir=./tensorboard/20250429-091731
+  job:
+    name: train
+    chdir: false
+    override_dirname: criterion=birdclef2025_categorical_cross_entropy,data=birdclef2025_15s,lr_scheduler=none,model=birdclef2025_convnext_tiny,optimizer=adamw_1e-4_decay_5e-2,preprocess.dump_format=birdclef2025,preprocess=birdclef2025,system=cuda,train.dataset.train.feature_dir=/kaggle/input/birdclef-2025,train.dataset.train.list_path=dump/birdclef2025_15s/list/train.txt,train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025,train.dataset.validation.list_path=dump/birdclef2025_15s/list/validation.txt,train.output.exp_dir=./exp/20250429-091731,train.output.tensorboard_dir=./tensorboard/20250429-091731,train.resume.continue_from=,train=birdclef2025_convnext_tiny
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.2'
+    cwd: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/ConvNeXtTiny
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /usr/local/lib/python3.10/dist-packages/audyn/configs
+      schema: file
+      provider: main
+    - path: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/ConvNeXtTiny/conf
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/log/20250429-091735
+    choices:
+      criterion: birdclef2025_categorical_cross_entropy
+      lr_scheduler: none
+      optimizer: adamw_1e-4_decay_5e-2
+      model: birdclef2025_convnext_tiny
+      test: default
+      test/dataloader: default
+      test/dataset: default
+      train: birdclef2025_convnext_tiny
+      train/record: default
+      train/clip_gradient: default
+      train/dataloader: default
+      train/dataset: birdclef2025_primary-label
+      data: birdclef2025_15s
+      preprocess: birdclef2025
+      system: cuda
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/log/20250429-091735/.hydra/overrides.yaml

@@ -0,0 +1,16 @@
+- system=cuda
+- preprocess=birdclef2025
+- data=birdclef2025_15s
+- train=birdclef2025_convnext_tiny
+- model=birdclef2025_convnext_tiny
+- optimizer=adamw_1e-4_decay_5e-2
+- lr_scheduler=none
+- criterion=birdclef2025_categorical_cross_entropy
+- preprocess.dump_format=birdclef2025
+- train.dataset.train.list_path=dump/birdclef2025_15s/list/train.txt
+- train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+- train.dataset.validation.list_path=dump/birdclef2025_15s/list/validation.txt
+- train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+- train.resume.continue_from=
+- train.output.exp_dir=./exp/20250429-091731
+- train.output.tensorboard_dir=./tensorboard/20250429-091731

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/log/20250429-091735/.hydra/resolved_config.yaml

@@ -0,0 +1,280 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: false
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: 2
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: 32000
+    hop_length: 1253
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 1253
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+        alpha: 0.4
+      num_workers: 2
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 1253
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: false
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+      num_workers: 2
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: log_melspectrogram
+      output: logit
+    validation:
+      input:
+        input: log_melspectrogram
+      output: logit
+    inference:
+      input:
+        input: log_melspectrogram
+      output: logit
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250429-091731
+    tensorboard_dir: ./tensorboard/20250429-091731
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ./exp/20250429-091731/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ./exp/20250429-091731/model/epoch{epoch}.pth
+      last:
+        path: ./exp/20250429-091731/model/last.pth
+      best_epoch:
+        path: ./exp/20250429-091731/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ./exp/inference
+    audio:
+      sample_rate: 32000
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+  ddp_kwargs: null
+model:
+  _target_: birdclef2025.models.ConvNeXtTiny
+  weights: IMAGENET1K_V1
+  num_classes: 206
+optimizer:
+  _target_: torch.optim.AdamW
+  lr: 0.0001
+  weight_decay: 0.05
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: label_index

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/model/best_epoch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c8c76a73f858b31c3e5e91cedb5bec0403da0ca62ebdd29cf069f83c4e130ff
+size 335992470

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-091731/model/last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eaca39c63d46df4018baadf070d222fdf55f1e611da5930fc7c02009442a642
+size 335953326

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/log/20250429-174603/.hydra/config.yaml

@@ -0,0 +1,223 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: null
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: null
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: ${..audio.sample_rate}
+    hop_length: 512
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: ${..train.audio_key}
+      sample_rate_key: ${..train.sample_rate_key}
+      label_name_key: ${..train.label_name_key}
+      filename_key: ${..train.filename_key}
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform: ${data.melspectrogram}
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: ${data.audio.sample_rate}
+          duration: ${data.audio.duration}
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+          target_shape: 256
+        melspectrogram_key: ${.composer.melspectrogram_key}
+        label_index_key: ${.composer.label_index_key}
+        alpha: 0.4
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: ${....train.collate_fn.composer._target_}
+          melspectrogram_transform: ${....train.collate_fn.composer.melspectrogram_transform}
+          audio_key: ${....train.collate_fn.composer.audio_key}
+          sample_rate_key: ${....train.collate_fn.composer.sample_rate_key}
+          label_name_key: ${....train.collate_fn.composer.label_name_key}
+          filename_key: ${....train.collate_fn.composer.filename_key}
+          waveform_key: ${....train.collate_fn.composer.waveform_key}
+          melspectrogram_key: ${....train.collate_fn.composer.melspectrogram_key}
+          label_index_key: ${....train.collate_fn.composer.label_index_key}
+          sample_rate: ${....train.collate_fn.composer.sample_rate}
+          duration: ${....train.collate_fn.composer.duration}
+          decode_audio_as_waveform: ${....train.collate_fn.composer.decode_audio_as_waveform}
+          decode_audio_as_monoral: ${....train.collate_fn.composer.decode_audio_as_monoral}
+          training: false
+          target_shape: ${....train.collate_fn.composer.target_shape}
+        melspectrogram_key: ${...train.collate_fn.composer.melspectrogram_key}
+        label_index_key: ${...train.collate_fn.composer.label_index_key}
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: ${....dataloader.train.collate_fn.composer.melspectrogram_key}
+      output: logit
+    validation: ${.train}
+    inference: ${.validation}
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250429-174600
+    tensorboard_dir: ./tensorboard/20250429-174600
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ${...exp_dir}/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ${...exp_dir}/model/epoch{epoch}.pth
+      last:
+        path: ${...exp_dir}/model/last.pth
+      best_epoch:
+        path: ${...exp_dir}/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ${.exp_dir}/inference
+    audio:
+      sample_rate: ${data.audio.sample_rate}
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+model:
+  _target_: birdclef2025.models.ConvNeXtTiny
+  weights: ${const:torchvision.models.ConvNeXt_Tiny_Weights.IMAGENET1K_V1}
+  num_classes: ${const:birdclef2025.utils.data.birdclef.num_birdclef2025_primary_labels}
+optimizer:
+  _target_: torch.optim.AdamW
+  lr: 0.0001
+  weight_decay: 0.05
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: ${train.dataloader.train.collate_fn.composer.label_index_key}

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/log/20250429-174603/.hydra/hydra.yaml

@@ -0,0 +1,189 @@
+hydra:
+  run:
+    dir: ./exp/20250429-174600/log/20250429-174603
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=./exp/20250429-174600/log/20250429-174603
+    - hydra.mode=RUN
+    task:
+    - system=cuda
+    - preprocess=birdclef2025
+    - data=birdclef2025_reshape_15s
+    - train=birdclef2025_reshape_convnext_tiny
+    - model=birdclef2025_convnext_tiny
+    - optimizer=adamw_1e-4_decay_5e-2
+    - lr_scheduler=none
+    - criterion=birdclef2025_categorical_cross_entropy
+    - preprocess.dump_format=birdclef2025
+    - train.dataset.train.list_path=dump/birdclef2025_reshape_15s/list/train.txt
+    - train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+    - train.dataset.validation.list_path=dump/birdclef2025_reshape_15s/list/validation.txt
+    - train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+    - train.resume.continue_from=
+    - train.output.exp_dir=./exp/20250429-174600
+    - train.output.tensorboard_dir=./tensorboard/20250429-174600
+  job:
+    name: train
+    chdir: false
+    override_dirname: criterion=birdclef2025_categorical_cross_entropy,data=birdclef2025_reshape_15s,lr_scheduler=none,model=birdclef2025_convnext_tiny,optimizer=adamw_1e-4_decay_5e-2,preprocess.dump_format=birdclef2025,preprocess=birdclef2025,system=cuda,train.dataset.train.feature_dir=/kaggle/input/birdclef-2025,train.dataset.train.list_path=dump/birdclef2025_reshape_15s/list/train.txt,train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025,train.dataset.validation.list_path=dump/birdclef2025_reshape_15s/list/validation.txt,train.output.exp_dir=./exp/20250429-174600,train.output.tensorboard_dir=./tensorboard/20250429-174600,train.resume.continue_from=,train=birdclef2025_reshape_convnext_tiny
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.2'
+    cwd: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/ConvNeXtTiny
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /usr/local/lib/python3.10/dist-packages/audyn/configs
+      schema: file
+      provider: main
+    - path: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/ConvNeXtTiny/conf
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/log/20250429-174603
+    choices:
+      criterion: birdclef2025_categorical_cross_entropy
+      lr_scheduler: none
+      optimizer: adamw_1e-4_decay_5e-2
+      model: birdclef2025_convnext_tiny
+      test: default
+      test/dataloader: default
+      test/dataset: default
+      train: birdclef2025_reshape_convnext_tiny
+      train/record: default
+      train/clip_gradient: default
+      train/dataloader: default
+      train/dataset: birdclef2025_primary-label
+      data: birdclef2025_reshape_15s
+      preprocess: birdclef2025
+      system: cuda
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/log/20250429-174603/.hydra/overrides.yaml

@@ -0,0 +1,16 @@
+- system=cuda
+- preprocess=birdclef2025
+- data=birdclef2025_reshape_15s
+- train=birdclef2025_reshape_convnext_tiny
+- model=birdclef2025_convnext_tiny
+- optimizer=adamw_1e-4_decay_5e-2
+- lr_scheduler=none
+- criterion=birdclef2025_categorical_cross_entropy
+- preprocess.dump_format=birdclef2025
+- train.dataset.train.list_path=dump/birdclef2025_reshape_15s/list/train.txt
+- train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+- train.dataset.validation.list_path=dump/birdclef2025_reshape_15s/list/validation.txt
+- train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+- train.resume.continue_from=
+- train.output.exp_dir=./exp/20250429-174600
+- train.output.tensorboard_dir=./tensorboard/20250429-174600

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/log/20250429-174603/.hydra/resolved_config.yaml

@@ -0,0 +1,282 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: false
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: 2
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: 32000
+    hop_length: 512
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 512
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+          target_shape: 256
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+        alpha: 0.4
+      num_workers: 2
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 512
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: false
+          target_shape: 256
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+      num_workers: 2
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: log_melspectrogram
+      output: logit
+    validation:
+      input:
+        input: log_melspectrogram
+      output: logit
+    inference:
+      input:
+        input: log_melspectrogram
+      output: logit
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250429-174600
+    tensorboard_dir: ./tensorboard/20250429-174600
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ./exp/20250429-174600/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ./exp/20250429-174600/model/epoch{epoch}.pth
+      last:
+        path: ./exp/20250429-174600/model/last.pth
+      best_epoch:
+        path: ./exp/20250429-174600/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ./exp/inference
+    audio:
+      sample_rate: 32000
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+  ddp_kwargs: null
+model:
+  _target_: birdclef2025.models.ConvNeXtTiny
+  weights: IMAGENET1K_V1
+  num_classes: 206
+optimizer:
+  _target_: torch.optim.AdamW
+  lr: 0.0001
+  weight_decay: 0.05
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: label_index

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/model/best_epoch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86838ad0507b599924a8eca92c0e9295cbd0fdd4858fd662153e1bdd4f710914
+size 335992534

recipes/BirdCLEF2025/ConvNeXtTiny/exp/20250429-174600/model/last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e05e2aaadfca28516aeda2d907ddcba8accdb5118e03ce38f63ff28395b8a155
+size 335953390

recipes/BirdCLEF2025/ConvNeXtTiny/tensorboard/20250429-091731/events.out.tfevents.1746371437.MacBook-Pro-2.local.7605.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:123ef3c4dcd02d0f1242a960e9c030ab8bddee904817d8759f7f1f2022f6dde1
+size 472294

recipes/BirdCLEF2025/ConvNeXtTiny/tensorboard/20250429-174600/events.out.tfevents.1746371438.MacBook-Pro-2.local.7605.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dc80cd9207cda3799c3d6d15f9d93126653ea9174346166bb454b8c60954e6e
+size 472294

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/README.md

@@ -0,0 +1,7 @@
+system=default_seed-0
+data=birdclef2025_5s
+train=birdclef2025_efficientnet_b0
+model=birdclef2025_efficientnet_b0
+criterion=birdclef2025_categorical_cross_entropy
+
+public_score: 0.479

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/log/20250427-104218/.hydra/config.yaml

@@ -0,0 +1,219 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: null
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: null
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 5
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: ${..audio.sample_rate}
+    hop_length: 1253
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_5s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_5s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: ${..train.audio_key}
+      sample_rate_key: ${..train.sample_rate_key}
+      label_name_key: ${..train.label_name_key}
+      filename_key: ${..train.filename_key}
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform: ${data.melspectrogram}
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: ${data.audio.sample_rate}
+          duration: ${data.audio.duration}
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+        melspectrogram_key: ${.composer.melspectrogram_key}
+        label_index_key: ${.composer.label_index_key}
+        alpha: 0.4
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: ${....train.collate_fn.composer._target_}
+          melspectrogram_transform: ${....train.collate_fn.composer.melspectrogram_transform}
+          audio_key: ${....train.collate_fn.composer.audio_key}
+          sample_rate_key: ${....train.collate_fn.composer.sample_rate_key}
+          label_name_key: ${....train.collate_fn.composer.label_name_key}
+          filename_key: ${....train.collate_fn.composer.filename_key}
+          waveform_key: ${....train.collate_fn.composer.waveform_key}
+          melspectrogram_key: ${....train.collate_fn.composer.melspectrogram_key}
+          label_index_key: ${....train.collate_fn.composer.label_index_key}
+          sample_rate: ${....train.collate_fn.composer.sample_rate}
+          duration: ${....train.collate_fn.composer.duration}
+          decode_audio_as_waveform: ${....train.collate_fn.composer.decode_audio_as_waveform}
+          decode_audio_as_monoral: ${....train.collate_fn.composer.decode_audio_as_monoral}
+          training: false
+        melspectrogram_key: ${...train.collate_fn.composer.melspectrogram_key}
+        label_index_key: ${...train.collate_fn.composer.label_index_key}
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: ${....dataloader.train.collate_fn.composer.melspectrogram_key}
+      output: logit
+    validation: ${.train}
+    inference: ${.validation}
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250427-104215
+    tensorboard_dir: ./tensorboard/20250427-104215
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ${...exp_dir}/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ${...exp_dir}/model/epoch{epoch}.pth
+      last:
+        path: ${...exp_dir}/model/last.pth
+      best_epoch:
+        path: ${...exp_dir}/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ${.exp_dir}/inference
+    audio:
+      sample_rate: ${data.audio.sample_rate}
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: ${const:torchvision.models.EfficientNet_B0_Weights.IMAGENET1K_V1}
+  num_classes: ${const:birdclef2025.utils.data.birdclef.num_birdclef2025_primary_labels}
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: ${train.dataloader.train.collate_fn.composer.label_index_key}

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/log/20250427-104218/.hydra/hydra.yaml

@@ -0,0 +1,189 @@
+hydra:
+  run:
+    dir: ./exp/20250427-104215/log/20250427-104218
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=./exp/20250427-104215/log/20250427-104218
+    - hydra.mode=RUN
+    task:
+    - system=cuda
+    - preprocess=birdclef2025
+    - data=birdclef2025_5s
+    - train=birdclef2025_efficientnet_b0
+    - model=birdclef2025_efficientnet_b0
+    - optimizer=adam
+    - lr_scheduler=none
+    - criterion=birdclef2025_categorical_cross_entropy
+    - preprocess.dump_format=birdclef2025
+    - train.dataset.train.list_path=dump/birdclef2025_5s/list/train.txt
+    - train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+    - train.dataset.validation.list_path=dump/birdclef2025_5s/list/validation.txt
+    - train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+    - train.resume.continue_from=
+    - train.output.exp_dir=./exp/20250427-104215
+    - train.output.tensorboard_dir=./tensorboard/20250427-104215
+  job:
+    name: train
+    chdir: false
+    override_dirname: criterion=birdclef2025_categorical_cross_entropy,data=birdclef2025_5s,lr_scheduler=none,model=birdclef2025_efficientnet_b0,optimizer=adam,preprocess.dump_format=birdclef2025,preprocess=birdclef2025,system=cuda,train.dataset.train.feature_dir=/kaggle/input/birdclef-2025,train.dataset.train.list_path=dump/birdclef2025_5s/list/train.txt,train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025,train.dataset.validation.list_path=dump/birdclef2025_5s/list/validation.txt,train.output.exp_dir=./exp/20250427-104215,train.output.tensorboard_dir=./tensorboard/20250427-104215,train.resume.continue_from=,train=birdclef2025_efficientnet_b0
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.2'
+    cwd: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /usr/local/lib/python3.10/dist-packages/audyn/configs
+      schema: file
+      provider: main
+    - path: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/conf
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/log/20250427-104218
+    choices:
+      criterion: birdclef2025_categorical_cross_entropy
+      lr_scheduler: none
+      optimizer: adam
+      model: birdclef2025_efficientnet_b0
+      test: default
+      test/dataloader: default
+      test/dataset: default
+      train: birdclef2025_efficientnet_b0
+      train/record: default
+      train/clip_gradient: default
+      train/dataloader: default
+      train/dataset: birdclef2025_primary-label
+      data: birdclef2025_5s
+      preprocess: birdclef2025
+      system: cuda
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/log/20250427-104218/.hydra/overrides.yaml

@@ -0,0 +1,16 @@
+- system=cuda
+- preprocess=birdclef2025
+- data=birdclef2025_5s
+- train=birdclef2025_efficientnet_b0
+- model=birdclef2025_efficientnet_b0
+- optimizer=adam
+- lr_scheduler=none
+- criterion=birdclef2025_categorical_cross_entropy
+- preprocess.dump_format=birdclef2025
+- train.dataset.train.list_path=dump/birdclef2025_5s/list/train.txt
+- train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+- train.dataset.validation.list_path=dump/birdclef2025_5s/list/validation.txt
+- train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+- train.resume.continue_from=
+- train.output.exp_dir=./exp/20250427-104215
+- train.output.tensorboard_dir=./tensorboard/20250427-104215

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/log/20250427-104218/.hydra/resolved_config.yaml

@@ -0,0 +1,278 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: false
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: 2
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 5
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: 32000
+    hop_length: 1253
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_5s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_5s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 1253
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 5
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+        alpha: 0.4
+      num_workers: 2
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 1253
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 5
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: false
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+      num_workers: 2
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: log_melspectrogram
+      output: logit
+    validation:
+      input:
+        input: log_melspectrogram
+      output: logit
+    inference:
+      input:
+        input: log_melspectrogram
+      output: logit
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250427-104215
+    tensorboard_dir: ./tensorboard/20250427-104215
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ./exp/20250427-104215/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ./exp/20250427-104215/model/epoch{epoch}.pth
+      last:
+        path: ./exp/20250427-104215/model/last.pth
+      best_epoch:
+        path: ./exp/20250427-104215/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ./exp/inference
+    audio:
+      sample_rate: 32000
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+  ddp_kwargs: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: IMAGENET1K_V1
+  num_classes: 206
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: label_index

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/model/best_epoch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bb7d8a078a4012e5dbad69974508f0e16913dddec612232a155ee5790805be2
+size 51764901

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-104215/model/last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:415711e50672e84febe12d7b656dd86d47ef46da0e9670c05bcab58a4e3b9a0d
+size 51716259

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/README.md

@@ -0,0 +1,7 @@
+system=default_seed-0
+data=birdclef2025_15s
+train=birdclef2025_efficientnet_b0
+model=birdclef2025_efficientnet_b0
+criterion=birdclef2025_categorical_cross_entropy
+
+public_score: 0.476

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/log/20250427-115147/.hydra/config.yaml

@@ -0,0 +1,219 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: null
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: null
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: ${..audio.sample_rate}
+    hop_length: 1253
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: ${..train.audio_key}
+      sample_rate_key: ${..train.sample_rate_key}
+      label_name_key: ${..train.label_name_key}
+      filename_key: ${..train.filename_key}
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform: ${data.melspectrogram}
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: ${data.audio.sample_rate}
+          duration: ${data.audio.duration}
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+        melspectrogram_key: ${.composer.melspectrogram_key}
+        label_index_key: ${.composer.label_index_key}
+        alpha: 0.4
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: ${....train.collate_fn.composer._target_}
+          melspectrogram_transform: ${....train.collate_fn.composer.melspectrogram_transform}
+          audio_key: ${....train.collate_fn.composer.audio_key}
+          sample_rate_key: ${....train.collate_fn.composer.sample_rate_key}
+          label_name_key: ${....train.collate_fn.composer.label_name_key}
+          filename_key: ${....train.collate_fn.composer.filename_key}
+          waveform_key: ${....train.collate_fn.composer.waveform_key}
+          melspectrogram_key: ${....train.collate_fn.composer.melspectrogram_key}
+          label_index_key: ${....train.collate_fn.composer.label_index_key}
+          sample_rate: ${....train.collate_fn.composer.sample_rate}
+          duration: ${....train.collate_fn.composer.duration}
+          decode_audio_as_waveform: ${....train.collate_fn.composer.decode_audio_as_waveform}
+          decode_audio_as_monoral: ${....train.collate_fn.composer.decode_audio_as_monoral}
+          training: false
+        melspectrogram_key: ${...train.collate_fn.composer.melspectrogram_key}
+        label_index_key: ${...train.collate_fn.composer.label_index_key}
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: ${....dataloader.train.collate_fn.composer.melspectrogram_key}
+      output: logit
+    validation: ${.train}
+    inference: ${.validation}
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250427-115144
+    tensorboard_dir: ./tensorboard/20250427-115144
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ${...exp_dir}/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ${...exp_dir}/model/epoch{epoch}.pth
+      last:
+        path: ${...exp_dir}/model/last.pth
+      best_epoch:
+        path: ${...exp_dir}/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ${.exp_dir}/inference
+    audio:
+      sample_rate: ${data.audio.sample_rate}
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: ${const:torchvision.models.EfficientNet_B0_Weights.IMAGENET1K_V1}
+  num_classes: ${const:birdclef2025.utils.data.birdclef.num_birdclef2025_primary_labels}
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: ${train.dataloader.train.collate_fn.composer.label_index_key}

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/log/20250427-115147/.hydra/hydra.yaml

@@ -0,0 +1,189 @@
+hydra:
+  run:
+    dir: ./exp/20250427-115144/log/20250427-115147
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=./exp/20250427-115144/log/20250427-115147
+    - hydra.mode=RUN
+    task:
+    - system=cuda
+    - preprocess=birdclef2025
+    - data=birdclef2025_15s
+    - train=birdclef2025_efficientnet_b0
+    - model=birdclef2025_efficientnet_b0
+    - optimizer=adam
+    - lr_scheduler=none
+    - criterion=birdclef2025_categorical_cross_entropy
+    - preprocess.dump_format=birdclef2025
+    - train.dataset.train.list_path=dump/birdclef2025_15s/list/train.txt
+    - train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+    - train.dataset.validation.list_path=dump/birdclef2025_15s/list/validation.txt
+    - train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+    - train.resume.continue_from=
+    - train.output.exp_dir=./exp/20250427-115144
+    - train.output.tensorboard_dir=./tensorboard/20250427-115144
+  job:
+    name: train
+    chdir: false
+    override_dirname: criterion=birdclef2025_categorical_cross_entropy,data=birdclef2025_15s,lr_scheduler=none,model=birdclef2025_efficientnet_b0,optimizer=adam,preprocess.dump_format=birdclef2025,preprocess=birdclef2025,system=cuda,train.dataset.train.feature_dir=/kaggle/input/birdclef-2025,train.dataset.train.list_path=dump/birdclef2025_15s/list/train.txt,train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025,train.dataset.validation.list_path=dump/birdclef2025_15s/list/validation.txt,train.output.exp_dir=./exp/20250427-115144,train.output.tensorboard_dir=./tensorboard/20250427-115144,train.resume.continue_from=,train=birdclef2025_efficientnet_b0
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.2'
+    cwd: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /usr/local/lib/python3.10/dist-packages/audyn/configs
+      schema: file
+      provider: main
+    - path: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/conf
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/log/20250427-115147
+    choices:
+      criterion: birdclef2025_categorical_cross_entropy
+      lr_scheduler: none
+      optimizer: adam
+      model: birdclef2025_efficientnet_b0
+      test: default
+      test/dataloader: default
+      test/dataset: default
+      train: birdclef2025_efficientnet_b0
+      train/record: default
+      train/clip_gradient: default
+      train/dataloader: default
+      train/dataset: birdclef2025_primary-label
+      data: birdclef2025_15s
+      preprocess: birdclef2025
+      system: cuda
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/log/20250427-115147/.hydra/overrides.yaml

@@ -0,0 +1,16 @@
+- system=cuda
+- preprocess=birdclef2025
+- data=birdclef2025_15s
+- train=birdclef2025_efficientnet_b0
+- model=birdclef2025_efficientnet_b0
+- optimizer=adam
+- lr_scheduler=none
+- criterion=birdclef2025_categorical_cross_entropy
+- preprocess.dump_format=birdclef2025
+- train.dataset.train.list_path=dump/birdclef2025_15s/list/train.txt
+- train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+- train.dataset.validation.list_path=dump/birdclef2025_15s/list/validation.txt
+- train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+- train.resume.continue_from=
+- train.output.exp_dir=./exp/20250427-115144
+- train.output.tensorboard_dir=./tensorboard/20250427-115144

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/log/20250427-115147/.hydra/resolved_config.yaml

@@ -0,0 +1,278 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: false
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: 2
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: 32000
+    hop_length: 1253
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 1253
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+        alpha: 0.4
+      num_workers: 2
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 1253
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: false
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+      num_workers: 2
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: log_melspectrogram
+      output: logit
+    validation:
+      input:
+        input: log_melspectrogram
+      output: logit
+    inference:
+      input:
+        input: log_melspectrogram
+      output: logit
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250427-115144
+    tensorboard_dir: ./tensorboard/20250427-115144
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ./exp/20250427-115144/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ./exp/20250427-115144/model/epoch{epoch}.pth
+      last:
+        path: ./exp/20250427-115144/model/last.pth
+      best_epoch:
+        path: ./exp/20250427-115144/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ./exp/inference
+    audio:
+      sample_rate: 32000
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+  ddp_kwargs: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: IMAGENET1K_V1
+  num_classes: 206
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: label_index

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/model/best_epoch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:887a11993732bc610e1fe1aba1f139ce608e5989ddba7e63575376045cab0511
+size 51764901

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-115144/model/last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4362e841a15c2fb999d73eb4ff6e50b84e2f5ea8822fa594771eab4043fa1d4
+size 51716259

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/README.md

@@ -0,0 +1,7 @@
+system=default_seed-0
+data=birdclef2025_reshape_15s
+train=birdclef2025_reshape_efficientnet_b0
+model=birdclef2025_efficientnet_b0
+criterion=birdclef2025_categorical_cross_entropy
+
+public_score: 0.519

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/log/20250427-133047/.hydra/config.yaml

@@ -0,0 +1,221 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: null
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: null
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: ${..audio.sample_rate}
+    hop_length: 512
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: ${..train.audio_key}
+      sample_rate_key: ${..train.sample_rate_key}
+      label_name_key: ${..train.label_name_key}
+      filename_key: ${..train.filename_key}
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform: ${data.melspectrogram}
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: ${data.audio.sample_rate}
+          duration: ${data.audio.duration}
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+          target_shape: 256
+        melspectrogram_key: ${.composer.melspectrogram_key}
+        label_index_key: ${.composer.label_index_key}
+        alpha: 0.4
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: ${....train.collate_fn.composer._target_}
+          melspectrogram_transform: ${....train.collate_fn.composer.melspectrogram_transform}
+          audio_key: ${....train.collate_fn.composer.audio_key}
+          sample_rate_key: ${....train.collate_fn.composer.sample_rate_key}
+          label_name_key: ${....train.collate_fn.composer.label_name_key}
+          filename_key: ${....train.collate_fn.composer.filename_key}
+          waveform_key: ${....train.collate_fn.composer.waveform_key}
+          melspectrogram_key: ${....train.collate_fn.composer.melspectrogram_key}
+          label_index_key: ${....train.collate_fn.composer.label_index_key}
+          sample_rate: ${....train.collate_fn.composer.sample_rate}
+          duration: ${....train.collate_fn.composer.duration}
+          decode_audio_as_waveform: ${....train.collate_fn.composer.decode_audio_as_waveform}
+          decode_audio_as_monoral: ${....train.collate_fn.composer.decode_audio_as_monoral}
+          training: false
+          target_shape: ${....train.collate_fn.composer.target_shape}
+        melspectrogram_key: ${...train.collate_fn.composer.melspectrogram_key}
+        label_index_key: ${...train.collate_fn.composer.label_index_key}
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: ${....dataloader.train.collate_fn.composer.melspectrogram_key}
+      output: logit
+    validation: ${.train}
+    inference: ${.validation}
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250427-133045
+    tensorboard_dir: ./tensorboard/20250427-133045
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ${...exp_dir}/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ${...exp_dir}/model/epoch{epoch}.pth
+      last:
+        path: ${...exp_dir}/model/last.pth
+      best_epoch:
+        path: ${...exp_dir}/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ${.exp_dir}/inference
+    audio:
+      sample_rate: ${data.audio.sample_rate}
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: ${const:torchvision.models.EfficientNet_B0_Weights.IMAGENET1K_V1}
+  num_classes: ${const:birdclef2025.utils.data.birdclef.num_birdclef2025_primary_labels}
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: ${train.dataloader.train.collate_fn.composer.label_index_key}

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/log/20250427-133047/.hydra/hydra.yaml

@@ -0,0 +1,189 @@
+hydra:
+  run:
+    dir: ./exp/20250427-133045/log/20250427-133047
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=./exp/20250427-133045/log/20250427-133047
+    - hydra.mode=RUN
+    task:
+    - system=cuda
+    - preprocess=birdclef2025
+    - data=birdclef2025_reshape_15s
+    - train=birdclef2025_reshape_efficientnet_b0
+    - model=birdclef2025_efficientnet_b0
+    - optimizer=adam
+    - lr_scheduler=none
+    - criterion=birdclef2025_categorical_cross_entropy
+    - preprocess.dump_format=birdclef2025
+    - train.dataset.train.list_path=dump/birdclef2025_reshape_15s/list/train.txt
+    - train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+    - train.dataset.validation.list_path=dump/birdclef2025_reshape_15s/list/validation.txt
+    - train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+    - train.resume.continue_from=
+    - train.output.exp_dir=./exp/20250427-133045
+    - train.output.tensorboard_dir=./tensorboard/20250427-133045
+  job:
+    name: train
+    chdir: false
+    override_dirname: criterion=birdclef2025_categorical_cross_entropy,data=birdclef2025_reshape_15s,lr_scheduler=none,model=birdclef2025_efficientnet_b0,optimizer=adam,preprocess.dump_format=birdclef2025,preprocess=birdclef2025,system=cuda,train.dataset.train.feature_dir=/kaggle/input/birdclef-2025,train.dataset.train.list_path=dump/birdclef2025_reshape_15s/list/train.txt,train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025,train.dataset.validation.list_path=dump/birdclef2025_reshape_15s/list/validation.txt,train.output.exp_dir=./exp/20250427-133045,train.output.tensorboard_dir=./tensorboard/20250427-133045,train.resume.continue_from=,train=birdclef2025_reshape_efficientnet_b0
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.2'
+    cwd: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /usr/local/lib/python3.10/dist-packages/audyn/configs
+      schema: file
+      provider: main
+    - path: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/conf
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/log/20250427-133047
+    choices:
+      criterion: birdclef2025_categorical_cross_entropy
+      lr_scheduler: none
+      optimizer: adam
+      model: birdclef2025_efficientnet_b0
+      test: default
+      test/dataloader: default
+      test/dataset: default
+      train: birdclef2025_reshape_efficientnet_b0
+      train/record: default
+      train/clip_gradient: default
+      train/dataloader: default
+      train/dataset: birdclef2025_primary-label
+      data: birdclef2025_reshape_15s
+      preprocess: birdclef2025
+      system: cuda
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/log/20250427-133047/.hydra/overrides.yaml

@@ -0,0 +1,16 @@
+- system=cuda
+- preprocess=birdclef2025
+- data=birdclef2025_reshape_15s
+- train=birdclef2025_reshape_efficientnet_b0
+- model=birdclef2025_efficientnet_b0
+- optimizer=adam
+- lr_scheduler=none
+- criterion=birdclef2025_categorical_cross_entropy
+- preprocess.dump_format=birdclef2025
+- train.dataset.train.list_path=dump/birdclef2025_reshape_15s/list/train.txt
+- train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+- train.dataset.validation.list_path=dump/birdclef2025_reshape_15s/list/validation.txt
+- train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+- train.resume.continue_from=
+- train.output.exp_dir=./exp/20250427-133045
+- train.output.tensorboard_dir=./tensorboard/20250427-133045

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/log/20250427-133047/.hydra/resolved_config.yaml

@@ -0,0 +1,280 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: false
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: 2
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 15
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: 32000
+    hop_length: 512
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_15s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 512
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+          target_shape: 256
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+        alpha: 0.4
+      num_workers: 2
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 512
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 15
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: false
+          target_shape: 256
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+      num_workers: 2
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: log_melspectrogram
+      output: logit
+    validation:
+      input:
+        input: log_melspectrogram
+      output: logit
+    inference:
+      input:
+        input: log_melspectrogram
+      output: logit
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250427-133045
+    tensorboard_dir: ./tensorboard/20250427-133045
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ./exp/20250427-133045/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ./exp/20250427-133045/model/epoch{epoch}.pth
+      last:
+        path: ./exp/20250427-133045/model/last.pth
+      best_epoch:
+        path: ./exp/20250427-133045/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ./exp/inference
+    audio:
+      sample_rate: 32000
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+  ddp_kwargs: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: IMAGENET1K_V1
+  num_classes: 206
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: label_index

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/model/best_epoch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dc01e358d6c9cdfe0d88db0cdef138aca9756e3751f8114a599b1a6f82f8ca3
+size 51764965

recipes/BirdCLEF2025/EfficientNetB0/exp/20250427-133045/model/last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab04bd9340de85728e8187eb6f92a7cd4fcc59c6492344fd410665968890aa90
+size 51716323

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/README.md

@@ -0,0 +1,7 @@
+system=default_seed-0
+data=birdclef2025_reshape_5s
+train=birdclef2025_reshape_efficientnet_b0
+model=birdclef2025_efficientnet_b0
+criterion=birdclef2025_categorical_cross_entropy
+
+public_score: 0.490

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/log/20250428-032425/.hydra/config.yaml

@@ -0,0 +1,221 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: null
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: null
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 5
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: ${..audio.sample_rate}
+    hop_length: 512
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_5s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_5s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: ${..train.audio_key}
+      sample_rate_key: ${..train.sample_rate_key}
+      label_name_key: ${..train.label_name_key}
+      filename_key: ${..train.filename_key}
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform: ${data.melspectrogram}
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: ${data.audio.sample_rate}
+          duration: ${data.audio.duration}
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+          target_shape: 256
+        melspectrogram_key: ${.composer.melspectrogram_key}
+        label_index_key: ${.composer.label_index_key}
+        alpha: 0.4
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: ${....train.collate_fn.composer._target_}
+          melspectrogram_transform: ${....train.collate_fn.composer.melspectrogram_transform}
+          audio_key: ${....train.collate_fn.composer.audio_key}
+          sample_rate_key: ${....train.collate_fn.composer.sample_rate_key}
+          label_name_key: ${....train.collate_fn.composer.label_name_key}
+          filename_key: ${....train.collate_fn.composer.filename_key}
+          waveform_key: ${....train.collate_fn.composer.waveform_key}
+          melspectrogram_key: ${....train.collate_fn.composer.melspectrogram_key}
+          label_index_key: ${....train.collate_fn.composer.label_index_key}
+          sample_rate: ${....train.collate_fn.composer.sample_rate}
+          duration: ${....train.collate_fn.composer.duration}
+          decode_audio_as_waveform: ${....train.collate_fn.composer.decode_audio_as_waveform}
+          decode_audio_as_monoral: ${....train.collate_fn.composer.decode_audio_as_monoral}
+          training: false
+          target_shape: ${....train.collate_fn.composer.target_shape}
+        melspectrogram_key: ${...train.collate_fn.composer.melspectrogram_key}
+        label_index_key: ${...train.collate_fn.composer.label_index_key}
+      num_workers: ${const:birdclef2025.utils.data.default_num_workers}
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: ${....dataloader.train.collate_fn.composer.melspectrogram_key}
+      output: logit
+    validation: ${.train}
+    inference: ${.validation}
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250428-032422
+    tensorboard_dir: ./tensorboard/20250428-032422
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ${...exp_dir}/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ${...exp_dir}/model/epoch{epoch}.pth
+      last:
+        path: ${...exp_dir}/model/last.pth
+      best_epoch:
+        path: ${...exp_dir}/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ${.exp_dir}/inference
+    audio:
+      sample_rate: ${data.audio.sample_rate}
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: ${const:torchvision.models.EfficientNet_B0_Weights.IMAGENET1K_V1}
+  num_classes: ${const:birdclef2025.utils.data.birdclef.num_birdclef2025_primary_labels}
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: ${train.dataloader.train.collate_fn.composer.label_index_key}

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/log/20250428-032425/.hydra/hydra.yaml

@@ -0,0 +1,189 @@
+hydra:
+  run:
+    dir: ./exp/20250428-032422/log/20250428-032425
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=./exp/20250428-032422/log/20250428-032425
+    - hydra.mode=RUN
+    task:
+    - system=cuda
+    - preprocess=birdclef2025
+    - data=birdclef2025_reshape_5s
+    - train=birdclef2025_reshape_efficientnet_b0
+    - model=birdclef2025_efficientnet_b0
+    - optimizer=adam
+    - lr_scheduler=none
+    - criterion=birdclef2025_categorical_cross_entropy
+    - preprocess.dump_format=birdclef2025
+    - train.dataset.train.list_path=dump/birdclef2025_reshape_5s/list/train.txt
+    - train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+    - train.dataset.validation.list_path=dump/birdclef2025_reshape_5s/list/validation.txt
+    - train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+    - train.resume.continue_from=
+    - train.output.exp_dir=./exp/20250428-032422
+    - train.output.tensorboard_dir=./tensorboard/20250428-032422
+  job:
+    name: train
+    chdir: false
+    override_dirname: criterion=birdclef2025_categorical_cross_entropy,data=birdclef2025_reshape_5s,lr_scheduler=none,model=birdclef2025_efficientnet_b0,optimizer=adam,preprocess.dump_format=birdclef2025,preprocess=birdclef2025,system=cuda,train.dataset.train.feature_dir=/kaggle/input/birdclef-2025,train.dataset.train.list_path=dump/birdclef2025_reshape_5s/list/train.txt,train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025,train.dataset.validation.list_path=dump/birdclef2025_reshape_5s/list/validation.txt,train.output.exp_dir=./exp/20250428-032422,train.output.tensorboard_dir=./tensorboard/20250428-032422,train.resume.continue_from=,train=birdclef2025_reshape_efficientnet_b0
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.2'
+    cwd: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /usr/local/lib/python3.10/dist-packages/audyn/configs
+      schema: file
+      provider: main
+    - path: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/conf
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /kaggle/working/BirdCLEF2025/recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/log/20250428-032425
+    choices:
+      criterion: birdclef2025_categorical_cross_entropy
+      lr_scheduler: none
+      optimizer: adam
+      model: birdclef2025_efficientnet_b0
+      test: default
+      test/dataloader: default
+      test/dataset: default
+      train: birdclef2025_reshape_efficientnet_b0
+      train/record: default
+      train/clip_gradient: default
+      train/dataloader: default
+      train/dataset: birdclef2025_primary-label
+      data: birdclef2025_reshape_5s
+      preprocess: birdclef2025
+      system: cuda
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/log/20250428-032425/.hydra/overrides.yaml

@@ -0,0 +1,16 @@
+- system=cuda
+- preprocess=birdclef2025
+- data=birdclef2025_reshape_5s
+- train=birdclef2025_reshape_efficientnet_b0
+- model=birdclef2025_efficientnet_b0
+- optimizer=adam
+- lr_scheduler=none
+- criterion=birdclef2025_categorical_cross_entropy
+- preprocess.dump_format=birdclef2025
+- train.dataset.train.list_path=dump/birdclef2025_reshape_5s/list/train.txt
+- train.dataset.train.feature_dir=/kaggle/input/birdclef-2025
+- train.dataset.validation.list_path=dump/birdclef2025_reshape_5s/list/validation.txt
+- train.dataset.validation.feature_dir=/kaggle/input/birdclef-2025
+- train.resume.continue_from=
+- train.output.exp_dir=./exp/20250428-032422
+- train.output.tensorboard_dir=./tensorboard/20250428-032422

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/log/20250428-032425/.hydra/resolved_config.yaml

@@ -0,0 +1,280 @@
+system:
+  seed: 0
+  distributed:
+    enable: null
+    nodes: null
+    nproc_per_node: null
+    backend: null
+    init_method: null
+    rdzv_id: null
+    rdzv_backend: null
+    rdzv_endpoint: null
+    max_restarts: null
+  cudnn:
+    benchmark: true
+    deterministic: false
+  amp:
+    enable: false
+    dtype: null
+  accelerator: cuda
+  compile:
+    enable: false
+    kwargs: null
+preprocess:
+  dump_format: birdclef2025
+  list_path: null
+  wav_dir: null
+  feature_dir: null
+  max_workers: 2
+  max_shard_size: 1000000000
+  vad:
+    raw_root: null
+    trimmed_root: null
+    threshold: null
+    min_duration: 15
+  csv_path: ???
+  submission_path: ???
+  audio_root: ???
+  subset: ???
+  train_ratio: 0.8
+data:
+  audio:
+    sample_rate: 32000
+    duration: 5
+  melspectrogram:
+    _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+    sample_rate: 32000
+    hop_length: 512
+    f_min: 20
+    f_max: 16000
+    pad: 0
+    n_mels: 128
+    window_fn:
+      _target_: torch.hann_window
+      _partial_: true
+    power: 1.0
+    normalized: false
+    wkwargs: null
+    center: true
+    pad_mode: constant
+    onesided: null
+    norm: slaney
+    mel_scale: slaney
+    take_log: true
+    freq_mask_param:
+    - 0.06
+    - 0.1
+    time_mask_param:
+    - 0.06
+    - 0.12
+    eps: null
+train:
+  dataset:
+    train:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_5s/list/train.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+    validation:
+      _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025PrimaryLabelDataset
+      list_path: dump/birdclef2025_reshape_5s/list/validation.txt
+      feature_dir: /kaggle/input/birdclef-2025
+      audio_key: audio
+      sample_rate_key: sample_rate
+      label_name_key: primary_label
+      filename_key: filename
+  dataloader:
+    train:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: true
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 512
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 5
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: true
+          target_shape: 256
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+        alpha: 0.4
+      num_workers: 2
+    validation:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 64
+      shuffle: false
+      collate_fn:
+        _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025BaselineValidationCollator
+        composer:
+          _target_: birdclef2025.utils.data.birdclef.BirdCLEF2025ReshapePrimaryLabelComposer
+          melspectrogram_transform:
+            _target_: birdclef2025.transforms.birdclef.BirdCLEF2024BaselineMelSpectrogram
+            sample_rate: 32000
+            hop_length: 512
+            f_min: 20
+            f_max: 16000
+            pad: 0
+            n_mels: 128
+            window_fn:
+              _target_: torch.hann_window
+              _partial_: true
+            power: 1.0
+            normalized: false
+            wkwargs: null
+            center: true
+            pad_mode: constant
+            onesided: null
+            norm: slaney
+            mel_scale: slaney
+            take_log: true
+            freq_mask_param:
+            - 0.06
+            - 0.1
+            time_mask_param:
+            - 0.06
+            - 0.12
+            eps: null
+          audio_key: audio
+          sample_rate_key: sample_rate
+          label_name_key: primary_label
+          filename_key: filename
+          waveform_key: waveform
+          melspectrogram_key: log_melspectrogram
+          label_index_key: label_index
+          sample_rate: 32000
+          duration: 5
+          decode_audio_as_waveform: true
+          decode_audio_as_monoral: true
+          training: false
+          target_shape: 256
+        melspectrogram_key: log_melspectrogram
+        label_index_key: label_index
+      num_workers: 2
+  clip_gradient: {}
+  record: {}
+  trainer:
+    _target_: birdclef2025.utils.driver.BaseTrainer
+  key_mapping:
+    train:
+      input:
+        input: log_melspectrogram
+      output: logit
+    validation:
+      input:
+        input: log_melspectrogram
+      output: logit
+    inference:
+      input:
+        input: log_melspectrogram
+      output: logit
+  ddp_kwargs: null
+  resume:
+    continue_from: ''
+  output:
+    exp_dir: ./exp/20250428-032422
+    tensorboard_dir: ./tensorboard/20250428-032422
+    save_checkpoint:
+      iteration:
+        every: 1000
+        path: ./exp/20250428-032422/model/iteration{iteration}.pth
+      epoch:
+        every: 1
+        path: ./exp/20250428-032422/model/epoch{epoch}.pth
+      last:
+        path: ./exp/20250428-032422/model/last.pth
+      best_epoch:
+        path: ./exp/20250428-032422/model/best_epoch.pth
+  steps:
+    epochs: 10
+    iterations: null
+    lr_scheduler: epoch
+test:
+  dataset:
+    test:
+      _target_: torch.utils.data.Dataset
+  dataloader:
+    test:
+      _target_: torch.utils.data.DataLoader
+      batch_size: 1
+      shuffle: false
+  key_mapping:
+    inference:
+      input: null
+      output: null
+      identifier: null
+  checkpoint: null
+  remove_weight_norm: null
+  output:
+    exp_dir: ./exp
+    inference_dir: ./exp/inference
+    audio:
+      sample_rate: 32000
+      key_mapping:
+        inference:
+          output: null
+          reference: null
+      transforms:
+        inference:
+          output: null
+          reference: null
+  ddp_kwargs: null
+model:
+  _target_: birdclef2025.models.EfficientNetB0
+  weights: IMAGENET1K_V1
+  num_classes: 206
+optimizer:
+  _target_: torch.optim.Adam
+lr_scheduler: {}
+criterion:
+  _target_: audyn.criterion.MultiCriteria
+  cross_entropy:
+    _target_: audyn.criterion.BaseCriterionWrapper
+    criterion:
+      _target_: torch.nn.CrossEntropyLoss
+      reduction: mean
+    weight: 1
+    key_mapping:
+      estimated:
+        input: logit
+      target:
+        target: label_index

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/model/best_epoch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58a676b0a4caf0423c81870af00dfc88979e4f52eb499e25bc835d3dcb3ae709
+size 51764965

recipes/BirdCLEF2025/EfficientNetB0/exp/20250428-032422/model/last.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5cf93364f17ff21617df6fee6c0ae03585377b160ee9e3b2e124ad19c20aef38
+size 51716323

recipes/BirdCLEF2025/EfficientNetB0/exp/20250429-075315/README.md

@@ -0,0 +1,7 @@
+system=default_seed-1
+data=birdclef2025_reshape_15s
+train=birdclef2025_reshape_efficientnet_b0
+model=birdclef2025_efficientnet_b0
+criterion=birdclef2025_categorical_cross_entropy
+
+public_score: 0.515