whisper-base-ga2en-v1.0

This model is a fine-tuned version of openai/whisper-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.5231
• Bleu: 19.21
• Chrf: 33.89
• Wer: 86.4025

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.03
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Chrf	Wer
0.5607	1.48	100	2.0068	13.89	29.73	102.1612
0.2227	2.96	200	2.1047	19.02	32.6	83.8361
0.0613	4.44	300	2.2670	20.37	34.38	82.9806
0.0491	5.93	400	2.2766	18.72	35.07	84.4665
0.0292	7.41	500	2.3870	19.19	33.57	85.2769
0.0307	8.89	600	2.4211	17.65	33.18	89.8244
0.0195	10.37	700	2.4980	19.56	33.37	83.3859
0.0165	11.85	800	2.4975	19.03	33.53	85.3219
0.0113	13.33	900	2.5300	18.93	33.95	85.9072
0.0097	14.81	1000	2.5231	19.21	33.89	86.4025

Framework versions

• Transformers 4.39.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2