nb-whisper-large-v0.8

This model is a fine-tuned version of NbAiLab/nb-whisper-large-v3-RC4 on the NbAiLab/ncc_speech_styling_v2 dataset. It achieves the following results on the evaluation set:

• step: 49999
• validation_nst_loss: 0.4309
• train_loss: 0.4828
• validation_nst_wer: 2.2211
• validation_nst_cer: 0.6758
• validation_nst_exact_wer: 2.7655
• validation_nst_exact_cer: 0.7592
• validation_clean_stortinget_no_loss: 0.7845
• validation_clean_stortinget_no_wer: 8.8323
• validation_clean_stortinget_no_cer: 5.6753
• validation_clean_stortinget_no_exact_wer: 11.6973
• validation_clean_stortinget_no_exact_cer: 6.1161

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 7e-05
• lr_scheduler_type: linear
• per_device_train_batch_size: 8
• total_train_batch_size_per_node: 32
• total_train_batch_size: 1024
• total_optimization_steps: 50,000
• starting_optimization_step: None
• finishing_optimization_step: 50,000
• num_train_dataset_workers: 32
• num_hosts: 32
• total_num_training_examples: 51,200,000
• steps_per_epoch: 7254
• num_beams: None
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.98
• adam_epsilon: 1e-06
• dropout: True
• bpe_dropout_probability: 0.2
• activation_dropout_probability: 0.1

Training results

step	validation_nst_loss	train_loss	validation_nst_wer	validation_nst_cer	validation_nst_exact_wer	validation_nst_exact_cer	validation_clean_stortinget_no_loss	validation_clean_stortinget_no_wer	validation_clean_stortinget_no_cer	validation_clean_stortinget_no_exact_wer	validation_clean_stortinget_no_exact_cer
0	0.4265	0.9701	2.1721	0.6246	2.7056	0.7070	0.6866	8.5836	5.4517	11.4126	5.8853
5000	0.4380	0.6065	2.5750	0.7495	3.0922	0.8251	0.6988	9.1284	5.8272	12.0840	6.2946
10000	0.4366	0.5640	2.3191	0.6852	2.8417	0.7647	0.7061	9.1378	5.7729	12.0270	6.2225
15000	0.4370	0.5506	2.3300	0.7066	2.9234	0.7976	0.7213	8.9673	5.6884	11.9511	6.1640
20000	0.4328	0.5284	2.3300	0.7019	2.8962	0.7885	0.7674	8.8915	5.6535	11.7922	6.1013
25000	0.4334	0.5133	2.3082	0.7010	2.9016	0.7903	0.7697	9.0194	5.7983	11.8468	6.2373
30000	0.4301	0.4996	2.1721	0.6674	2.6948	0.7464	0.7732	8.9223	5.7229	11.8349	6.1726
35000	0.4310	0.4957	2.2592	0.6926	2.8472	0.7830	0.7882	8.9744	5.7804	11.8871	6.2323
40000	0.4301	0.4999	2.1939	0.6647	2.7165	0.7436	0.7899	8.8868	5.6412	11.7708	6.0880
45000	0.4306	0.5049	2.2320	0.6768	2.7819	0.7628	0.7766	8.8252	5.6686	11.6902	6.1087
49999	0.4309	0.4828	2.2211	0.6758	2.7655	0.7592
49999	0.7845	0.4828	8.8323	5.6753	11.6973	6.1161

Framework versions

• Transformers 4.36.2
• Datasets 2.16.0
• Tokenizers 0.15.0