whisper-large-v3-genbed-f-model

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: 0.5346
• Wer: 33.8051

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: 1.75e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 30000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.0784	0.6605	250	0.5140	48.6274
0.4405	1.3210	500	0.4665	40.7746
0.3641	1.9815	750	0.4253	37.1462
0.215	2.6420	1000	0.4413	35.1990
0.1871	3.3025	1250	0.4725	37.4548
0.1425	3.9630	1500	0.4407	34.2520
0.0918	4.6235	1750	0.4618	33.9860
0.0821	5.2840	2000	0.4980	33.8689
0.0665	5.9445	2250	0.5042	32.3367
0.048	6.6050	2500	0.4927	33.9860
0.0441	7.2655	2750	0.5449	32.0919
0.0387	7.9260	3000	0.5235	31.6876
0.0307	8.5865	3250	0.5227	31.7408
0.0282	9.2470	3500	0.5682	32.3792
0.0288	9.9075	3750	0.5346	33.8051

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1