Pageee/DistilFT-English-10ha

DistilFT-English-10h

This model is a fine-tuned version of distil-small.en on the librispeech dataset. It achieves the following results on the evaluation set:

• Loss: 0.2318
• Wer: 4.4905

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-07
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 300
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.651	0.5556	100	0.9641	3.4754
0.5006	1.1111	200	0.7651	3.5039
0.3531	1.6667	300	0.5188	3.5121
0.2176	2.2222	400	0.3514	4.0258
0.1834	2.7778	500	0.2878	4.3132
0.1587	3.3333	600	0.2589	4.4049
0.1553	3.8889	700	0.2447	4.5007
0.1566	4.4444	800	0.2370	4.5007
0.1226	5.0	900	0.2332	4.5048
0.1533	5.5556	1000	0.2318	4.4905

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1