whisper-large-v3-cv-capes-filtered-pt

This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1440

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: 5e-06
• train_batch_size: 128
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
0.1991	0.2833	50	0.1383
0.1258	0.5666	100	0.1127
0.1117	0.8499	150	0.1086
0.0848	1.1303	200	0.1065
0.0874	1.4136	250	0.1060
0.0807	1.6969	300	0.1055
0.0793	1.9802	350	0.1055
0.0598	2.2606	400	0.1111
0.0591	2.5439	450	0.1156
0.0581	2.8272	500	0.1117
0.0447	3.1076	550	0.1226
0.0448	3.3909	600	0.1210
0.0478	3.6742	650	0.1230
0.0426	3.9575	700	0.1225
0.0326	4.2380	750	0.1339
0.0339	4.5212	800	0.1360
0.0356	4.8045	850	0.1338
0.0283	5.0850	900	0.1430
0.0298	5.3683	950	0.1449
0.0267	5.6516	1000	0.1440

Framework versions

• Transformers 4.50.2
• Pytorch 2.5.1+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2