whisper-adult-100

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

• Loss: 0.4262
• Wer: 21.1427

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: 48
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 57
• num_epochs: 3.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.102	0.1316	25	1.2196	38.0649
1.0942	0.2632	50	1.1864	37.6388
1.0216	0.3947	75	1.1233	36.4344
0.8959	0.5263	100	1.0434	35.3239
0.8829	0.6579	125	0.9577	34.7938
0.8066	0.7895	150	0.8676	32.4320
0.7629	0.9211	175	0.7698	33.1734
0.6601	1.0526	200	0.6740	31.0766
0.6208	1.1842	225	0.6014	28.5000
0.5623	1.3158	250	0.5413	25.5443
0.5087	1.4474	275	0.4998	24.2158
0.5239	1.5789	300	0.4764	23.2798
0.4666	1.7105	325	0.4612	23.8702
0.4788	1.8421	350	0.4505	22.8336
0.5029	1.9737	375	0.4433	22.8101
0.4644	2.1053	400	0.4384	22.0217
0.4877	2.2368	425	0.4346	22.2230
0.4624	2.3684	450	0.4318	22.1492
0.479	2.5	475	0.4295	21.6728
0.4797	2.6316	500	0.4278	21.6627
0.4661	2.7632	525	0.4267	21.1662
0.4411	2.8947	550	0.4262	21.1427

Framework versions

• PEFT 0.16.0
• Transformers 4.52.0
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.2