sanchit-gandhi/distil-large-v3-hi-ft-frozen-encoder

distil-whisper/distil-large-v3

This model is a fine-tuned version of distil-whisper/distil-large-v3 on the mozilla-foundation/common_voice_16_1 hi dataset. It achieves the following results on the evaluation set:

• Loss: 0.3749
• Wer: 0.2664

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• 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
0.1035	4.5	1000	0.3015	0.3250
0.0165	9.01	2000	0.3496	0.3007
0.0022	13.51	3000	0.3649	0.2786
0.0011	18.02	4000	0.3700	0.2681
0.0003	22.52	5000	0.3749	0.2664

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.1