parambharat/whisper-tiny-ml

Whisper Tiny ml - Bharat Ramanathan

This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1286
• Wer: 106.9296

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: 32
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5755	4.02	500	0.4241	81.2652
0.4182	9.01	1000	0.3245	72.7494
0.3387	14.01	1500	0.2914	67.2749
0.2923	19.0	2000	0.2745	60.3406
0.2596	24.0	2500	0.2645	58.2725
0.2356	28.02	3000	0.2629	60.3406
0.2167	33.01	3500	0.2647	59.9757
0.2039	4.02	4000	0.2617	58.2725
0.1938	9.01	4500	0.2644	58.2725
0.1858	14.01	5000	0.2636	58.7591

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2