whisper-large-v2-yodas-2

This model is a fine-tuned version of openai/whisper-large-v2 on the fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 0.4643

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.8132	0.2473	1000	0.3183
0.7788	0.4947	2000	0.3530
0.7566	0.7420	3000	0.3373
0.7586	0.9894	4000	0.3444
0.5781	1.2367	5000	0.3332
0.5901	1.4840	6000	0.3637
0.5837	1.7314	7000	0.3439
0.5662	1.9787	8000	0.3573
0.3619	2.2261	9000	0.3648
0.3695	2.4734	10000	0.3754
0.3713	2.7208	11000	0.3572
0.3804	2.9681	12000	0.3732
0.2004	3.2154	13000	0.4276
0.1987	3.4628	14000	0.4003
0.2006	3.7101	15000	0.3896
0.2077	3.9575	16000	0.3951
0.0913	4.2048	17000	0.4249
0.0901	4.4521	18000	0.4335
0.0885	4.6995	19000	0.4430
0.0875	4.9468	20000	0.4345
0.032	5.1942	21000	0.4428
0.0345	5.4415	22000	0.4609
0.0343	5.6888	23000	0.4630
0.0324	5.9362	24000	0.4643

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.18.0
• Tokenizers 0.19.1