ESPnet
audio
classification
Shikhar Bharadwaj commited on
Commit
557ac05
·
1 Parent(s): 6241f6c

Update model

Browse files
Files changed (19) hide show
  1. README.md +255 -0
  2. meta.yaml +8 -0
  3. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/data/beans_humbugdb/token_list +16 -0
  4. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/201epoch.pth +3 -0
  5. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/RESULTS.md +16 -0
  6. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/config.yaml +203 -0
  7. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/acc.png +0 -0
  8. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/backward_time.png +0 -0
  9. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/clip.png +0 -0
  10. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/forward_time.png +0 -0
  11. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/gpu_max_cached_mem_GB.png +0 -0
  12. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/grad_norm.png +0 -0
  13. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/iter_time.png +0 -0
  14. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/loss.png +0 -0
  15. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/loss_scale.png +0 -0
  16. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/macro_precision.png +0 -0
  17. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/optim0_lr0.png +0 -0
  18. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/optim_step_time.png +0 -0
  19. work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/train_time.png +0 -0
README.md ADDED
@@ -0,0 +1,255 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - espnet
4
+ - audio
5
+ - classification
6
+ datasets:
7
+ - beans
8
+ license: cc-by-4.0
9
+ ---
10
+
11
+ ## ESPnet2 CLS model
12
+
13
+ ### `espnet/OpenBEATS-Large-i3-humbugdb`
14
+
15
+ This model was trained by Shikhar Bharadwaj using beans recipe in [espnet](https://github.com/espnet/espnet/).
16
+
17
+ ## CLS config
18
+
19
+ <details><summary>expand</summary>
20
+
21
+ ```
22
+ config: /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/earlarge3/conf/ear_large/beans_humbugdb.yaml
23
+ print_config: false
24
+ log_level: INFO
25
+ drop_last_iter: false
26
+ dry_run: false
27
+ iterator_type: sequence
28
+ valid_iterator_type: null
29
+ output_dir: /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3
30
+ ngpu: 1
31
+ seed: 0
32
+ num_workers: 2
33
+ num_att_plot: 0
34
+ dist_backend: nccl
35
+ dist_init_method: env://
36
+ dist_world_size: null
37
+ dist_rank: null
38
+ local_rank: 0
39
+ dist_master_addr: null
40
+ dist_master_port: null
41
+ dist_launcher: null
42
+ multiprocessing_distributed: false
43
+ unused_parameters: true
44
+ sharded_ddp: false
45
+ use_deepspeed: false
46
+ deepspeed_config: null
47
+ gradient_as_bucket_view: true
48
+ ddp_comm_hook: null
49
+ cudnn_enabled: true
50
+ cudnn_benchmark: false
51
+ cudnn_deterministic: true
52
+ use_tf32: false
53
+ collect_stats: false
54
+ write_collected_feats: false
55
+ max_epoch: 250
56
+ patience: null
57
+ val_scheduler_criterion:
58
+ - valid
59
+ - loss
60
+ early_stopping_criterion:
61
+ - valid
62
+ - loss
63
+ - min
64
+ best_model_criterion:
65
+ - - valid
66
+ - acc
67
+ - max
68
+ keep_nbest_models: 1
69
+ nbest_averaging_interval: 0
70
+ grad_clip: 1
71
+ grad_clip_type: 2.0
72
+ grad_noise: false
73
+ accum_grad: 1
74
+ no_forward_run: false
75
+ resume: true
76
+ train_dtype: float32
77
+ use_amp: false
78
+ log_interval: null
79
+ use_matplotlib: true
80
+ use_tensorboard: true
81
+ create_graph_in_tensorboard: false
82
+ use_wandb: true
83
+ wandb_project: audioverse
84
+ wandb_id: null
85
+ wandb_entity: shikhar
86
+ wandb_name: beans_humbugdb.earlarge3
87
+ wandb_model_log_interval: -1
88
+ detect_anomaly: false
89
+ use_adapter: false
90
+ adapter: lora
91
+ save_strategy: all
92
+ adapter_conf: {}
93
+ pretrain_path: null
94
+ init_param: []
95
+ ignore_init_mismatch: false
96
+ freeze_param: []
97
+ num_iters_per_epoch: null
98
+ batch_size: 16
99
+ valid_batch_size: 16
100
+ batch_bins: 1000000
101
+ valid_batch_bins: null
102
+ category_sample_size: 10
103
+ train_shape_file:
104
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/train/speech_shape
105
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/train/label_shape
106
+ valid_shape_file:
107
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/valid/speech_shape
108
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/valid/label_shape
109
+ batch_type: folded
110
+ valid_batch_type: null
111
+ fold_length:
112
+ - 160000
113
+ - 5
114
+ sort_in_batch: descending
115
+ shuffle_within_batch: false
116
+ sort_batch: descending
117
+ multiple_iterator: false
118
+ utt2weight_file: null
119
+ chunk_length: 500
120
+ chunk_shift_ratio: 0.5
121
+ num_cache_chunks: 1024
122
+ chunk_excluded_key_prefixes: []
123
+ chunk_default_fs: null
124
+ chunk_max_abs_length: null
125
+ chunk_discard_short_samples: true
126
+ train_data_path_and_name_and_type:
127
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.train/wav.scp
128
+ - speech
129
+ - sound
130
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.train/text
131
+ - label
132
+ - text
133
+ valid_data_path_and_name_and_type:
134
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.dev/wav.scp
135
+ - speech
136
+ - sound
137
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.dev/text
138
+ - label
139
+ - text
140
+ multi_task_dataset: false
141
+ allow_variable_data_keys: false
142
+ max_cache_size: 0.0
143
+ max_cache_fd: 32
144
+ allow_multi_rates: false
145
+ valid_max_cache_size: null
146
+ exclude_weight_decay: false
147
+ exclude_weight_decay_conf: {}
148
+ optim: adamw
149
+ optim_conf:
150
+ lr: 3.0e-05
151
+ weight_decay: 0.01
152
+ betas:
153
+ - 0.9
154
+ - 0.98
155
+ scheduler: cosineannealingwarmuprestarts
156
+ scheduler_conf:
157
+ first_cycle_steps: 95000
158
+ warmup_steps: 8000
159
+ max_lr: 3.0e-05
160
+ min_lr: 5.0e-06
161
+ lightning_conf: {}
162
+ token_list:
163
+ - non-mosquito
164
+ - an_arabiensis
165
+ - an_gambiae_ss
166
+ - others
167
+ - culex_quinquefasciatus
168
+ - culex_pipiens_complex
169
+ - an_funestus_ss
170
+ - an_squamosus
171
+ - ma_uniformis
172
+ - an_dirus
173
+ - an_harrisoni
174
+ - ae_aegypti
175
+ - an_maculatus
176
+ - an_funestus_sl
177
+ - <blank>
178
+ - <unk>
179
+ text_token_list: null
180
+ text_bpemodel: null
181
+ init: xavier_normal
182
+ input_size: 1
183
+ use_preprocessor: true
184
+ frontend: null
185
+ frontend_conf: {}
186
+ specaug: null
187
+ specaug_conf: {}
188
+ normalize: null
189
+ normalize_conf: {}
190
+ preencoder: null
191
+ preencoder_conf: {}
192
+ encoder: beats
193
+ encoder_conf:
194
+ beats_ckpt_path: /work/nvme/bbjs/sbharadwaj/7Msounds/exp/beats_iter2_large2.tune_lr1.0e-4_warmup40000_bins1600000_totalsteps400000/epoch_latest.pt
195
+ beats_config:
196
+ layer_wise_gradient_decay_ratio: 0.3
197
+ encoder_layerdrop: 0.1
198
+ dropout: 0.0
199
+ use_weighted_representation: false
200
+ specaug_config:
201
+ apply_time_warp: true
202
+ apply_freq_mask: false
203
+ apply_time_mask: true
204
+ time_mask_width_ratio_range:
205
+ - 0
206
+ - 0.06
207
+ num_time_mask: 1
208
+ roll_augment: true
209
+ roll_interval: 1
210
+ text_encoder: null
211
+ text_encoder_conf: {}
212
+ embedding_fusion: null
213
+ embedding_fusion_conf: {}
214
+ decoder: linear
215
+ decoder_conf: {}
216
+ model: espnet
217
+ model_conf:
218
+ classification_type: multi-class
219
+ lsm_weight: 0.1
220
+ required:
221
+ - output_dir
222
+ - token_list
223
+ version: '202412'
224
+ distributed: false
225
+ ```
226
+
227
+ </details>
228
+
229
+ ### Citations
230
+
231
+ ```BibTex
232
+
233
+ @article{bharadwaj2025openbeats,
234
+ title={OpenBEATs: A Fully Open-Source General-Purpose Audio Encoder},
235
+ author={Bharadwaj, Shikhar and Cornell, Samuele and Choi, Kwanghee and Fukayama, Satoru and Shim, Hye-jin and Deshmukh, Soham and Watanabe, Shinji},
236
+ journal={arXiv preprint arXiv:2507.14129},
237
+ year={2025}
238
+ }
239
+
240
+ @inproceedings{watanabe2018espnet,
241
+ author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
242
+ title={{ESPnet}: End-to-End Speech Processing Toolkit},
243
+ year={2018},
244
+ booktitle={Proceedings of Interspeech},
245
+ pages={2207--2211},
246
+ doi={10.21437/Interspeech.2018-1456},
247
+ url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
248
+ }
249
+
250
+
251
+
252
+
253
+
254
+
255
+ ```
meta.yaml ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ espnet: '202503'
2
+ files:
3
+ classification_model_file: /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/201epoch.pth
4
+ python: "3.9.18 | packaged by conda-forge | (main, Dec 23 2023, 17:20:25) \n[GCC 12.3.0]"
5
+ timestamp: 1763331518.780911
6
+ torch: 2.1.2
7
+ yaml_files:
8
+ classification_train_config: /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/config.yaml
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/data/beans_humbugdb/token_list ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ non-mosquito
2
+ an_arabiensis
3
+ an_gambiae_ss
4
+ others
5
+ culex_quinquefasciatus
6
+ culex_pipiens_complex
7
+ an_funestus_ss
8
+ an_squamosus
9
+ ma_uniformis
10
+ an_dirus
11
+ an_harrisoni
12
+ ae_aegypti
13
+ an_maculatus
14
+ an_funestus_sl
15
+ <blank>
16
+ <unk>
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/201epoch.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:732984d44eff72f9d945abce581c987a25e46bb9353b5c4f19aa3461fbfef169
3
+ size 1245753311
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/RESULTS.md ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <!-- Generated by scripts/utils/show_cls_result.sh -->
2
+ # RESULTS
3
+ ## Environments
4
+ - date: `Sat Mar 22 07:56:50 CDT 2025`
5
+ - python version: `3.9.18 | packaged by conda-forge | (main, Dec 23 2023, 17:20:25) [GCC 12.3.0]`
6
+ - espnet version: `espnet 202412`
7
+ - pytorch version: `pytorch 2.6.0.dev20241210+cu124`
8
+ - Git hash: `ee8dd3d5da745a2c08c2bd6518bc0ba41ba5b224`
9
+ - Commit date: `Thu Mar 20 16:45:17 2025 -0500`
10
+
11
+ ## cls_earlarge3
12
+ |Split|mean_acc|mAP|mean_auc|n_labels|n_instances|
13
+ |---|---|---|---|---|---|
14
+ cls_humbugdb.dev|87.89|69.50|95.21|14.00|1833.00
15
+ cls_humbugdb.test|87.41|68.79|95.29|14.00|1859.00
16
+
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/config.yaml ADDED
@@ -0,0 +1,203 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ config: /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/earlarge3/conf/ear_large/beans_humbugdb.yaml
2
+ print_config: false
3
+ log_level: INFO
4
+ drop_last_iter: false
5
+ dry_run: false
6
+ iterator_type: sequence
7
+ valid_iterator_type: null
8
+ output_dir: /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3
9
+ ngpu: 1
10
+ seed: 0
11
+ num_workers: 2
12
+ num_att_plot: 0
13
+ dist_backend: nccl
14
+ dist_init_method: env://
15
+ dist_world_size: null
16
+ dist_rank: null
17
+ local_rank: 0
18
+ dist_master_addr: null
19
+ dist_master_port: null
20
+ dist_launcher: null
21
+ multiprocessing_distributed: false
22
+ unused_parameters: true
23
+ sharded_ddp: false
24
+ use_deepspeed: false
25
+ deepspeed_config: null
26
+ gradient_as_bucket_view: true
27
+ ddp_comm_hook: null
28
+ cudnn_enabled: true
29
+ cudnn_benchmark: false
30
+ cudnn_deterministic: true
31
+ use_tf32: false
32
+ collect_stats: false
33
+ write_collected_feats: false
34
+ max_epoch: 250
35
+ patience: null
36
+ val_scheduler_criterion:
37
+ - valid
38
+ - loss
39
+ early_stopping_criterion:
40
+ - valid
41
+ - loss
42
+ - min
43
+ best_model_criterion:
44
+ - - valid
45
+ - acc
46
+ - max
47
+ keep_nbest_models: 1
48
+ nbest_averaging_interval: 0
49
+ grad_clip: 1
50
+ grad_clip_type: 2.0
51
+ grad_noise: false
52
+ accum_grad: 1
53
+ no_forward_run: false
54
+ resume: true
55
+ train_dtype: float32
56
+ use_amp: false
57
+ log_interval: null
58
+ use_matplotlib: true
59
+ use_tensorboard: true
60
+ create_graph_in_tensorboard: false
61
+ use_wandb: true
62
+ wandb_project: audioverse
63
+ wandb_id: null
64
+ wandb_entity: shikhar
65
+ wandb_name: beans_humbugdb.earlarge3
66
+ wandb_model_log_interval: -1
67
+ detect_anomaly: false
68
+ use_adapter: false
69
+ adapter: lora
70
+ save_strategy: all
71
+ adapter_conf: {}
72
+ pretrain_path: null
73
+ init_param: []
74
+ ignore_init_mismatch: false
75
+ freeze_param: []
76
+ num_iters_per_epoch: null
77
+ batch_size: 16
78
+ valid_batch_size: 16
79
+ batch_bins: 1000000
80
+ valid_batch_bins: null
81
+ category_sample_size: 10
82
+ train_shape_file:
83
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/train/speech_shape
84
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/train/label_shape
85
+ valid_shape_file:
86
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/valid/speech_shape
87
+ - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_stats_16k/valid/label_shape
88
+ batch_type: folded
89
+ valid_batch_type: null
90
+ fold_length:
91
+ - 160000
92
+ - 5
93
+ sort_in_batch: descending
94
+ shuffle_within_batch: false
95
+ sort_batch: descending
96
+ multiple_iterator: false
97
+ utt2weight_file: null
98
+ chunk_length: 500
99
+ chunk_shift_ratio: 0.5
100
+ num_cache_chunks: 1024
101
+ chunk_excluded_key_prefixes: []
102
+ chunk_default_fs: null
103
+ chunk_max_abs_length: null
104
+ chunk_discard_short_samples: true
105
+ train_data_path_and_name_and_type:
106
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.train/wav.scp
107
+ - speech
108
+ - sound
109
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.train/text
110
+ - label
111
+ - text
112
+ valid_data_path_and_name_and_type:
113
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.dev/wav.scp
114
+ - speech
115
+ - sound
116
+ - - /work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/dump/beans_humbugdb/humbugdb.dev/text
117
+ - label
118
+ - text
119
+ multi_task_dataset: false
120
+ allow_variable_data_keys: false
121
+ max_cache_size: 0.0
122
+ max_cache_fd: 32
123
+ allow_multi_rates: false
124
+ valid_max_cache_size: null
125
+ exclude_weight_decay: false
126
+ exclude_weight_decay_conf: {}
127
+ optim: adamw
128
+ optim_conf:
129
+ lr: 3.0e-05
130
+ weight_decay: 0.01
131
+ betas:
132
+ - 0.9
133
+ - 0.98
134
+ scheduler: cosineannealingwarmuprestarts
135
+ scheduler_conf:
136
+ first_cycle_steps: 95000
137
+ warmup_steps: 8000
138
+ max_lr: 3.0e-05
139
+ min_lr: 5.0e-06
140
+ lightning_conf: {}
141
+ token_list:
142
+ - non-mosquito
143
+ - an_arabiensis
144
+ - an_gambiae_ss
145
+ - others
146
+ - culex_quinquefasciatus
147
+ - culex_pipiens_complex
148
+ - an_funestus_ss
149
+ - an_squamosus
150
+ - ma_uniformis
151
+ - an_dirus
152
+ - an_harrisoni
153
+ - ae_aegypti
154
+ - an_maculatus
155
+ - an_funestus_sl
156
+ - <blank>
157
+ - <unk>
158
+ text_token_list: null
159
+ text_bpemodel: null
160
+ init: xavier_normal
161
+ input_size: 1
162
+ use_preprocessor: true
163
+ frontend: null
164
+ frontend_conf: {}
165
+ specaug: null
166
+ specaug_conf: {}
167
+ normalize: null
168
+ normalize_conf: {}
169
+ preencoder: null
170
+ preencoder_conf: {}
171
+ encoder: beats
172
+ encoder_conf:
173
+ beats_ckpt_path: /work/nvme/bbjs/sbharadwaj/7Msounds/exp/beats_iter2_large2.tune_lr1.0e-4_warmup40000_bins1600000_totalsteps400000/epoch_latest.pt
174
+ beats_config:
175
+ layer_wise_gradient_decay_ratio: 0.3
176
+ encoder_layerdrop: 0.1
177
+ dropout: 0.0
178
+ use_weighted_representation: false
179
+ specaug_config:
180
+ apply_time_warp: true
181
+ apply_freq_mask: false
182
+ apply_time_mask: true
183
+ time_mask_width_ratio_range:
184
+ - 0
185
+ - 0.06
186
+ num_time_mask: 1
187
+ roll_augment: true
188
+ roll_interval: 1
189
+ text_encoder: null
190
+ text_encoder_conf: {}
191
+ embedding_fusion: null
192
+ embedding_fusion_conf: {}
193
+ decoder: linear
194
+ decoder_conf: {}
195
+ model: espnet
196
+ model_conf:
197
+ classification_type: multi-class
198
+ lsm_weight: 0.1
199
+ required:
200
+ - output_dir
201
+ - token_list
202
+ version: '202412'
203
+ distributed: false
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/acc.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/backward_time.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/clip.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/forward_time.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/gpu_max_cached_mem_GB.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/grad_norm.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/iter_time.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/loss.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/loss_scale.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/macro_precision.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/optim0_lr0.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/optim_step_time.png ADDED
work/nvme/bbjs/sbharadwaj/espnet/egs2/audioverse/v1/exp/beans_humbugdb/cls_earlarge3/images/train_time.png ADDED