whisper-medium-anime-5k

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

• Loss: 0.2012

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: 8
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.3
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.3486	0.1082	200	1.9858
1.2438	0.2165	400	0.7836
0.3552	0.3247	600	0.2941
0.2846	0.4329	800	0.2758
0.2712	0.5411	1000	0.2623
0.2602	0.6494	1200	0.2537
0.253	0.7576	1400	0.2450
0.2397	0.8658	1600	0.2387
0.2364	0.9740	1800	0.2348
0.2238	1.0823	2000	0.2292
0.2147	1.1905	2200	0.2262
0.2129	1.2987	2400	0.2220
0.2123	1.4069	2600	0.2189
0.2118	1.5152	2800	0.2165
0.2067	1.6234	3000	0.2140
0.2077	1.7316	3200	0.2123
0.2012	1.8398	3400	0.2089
0.1987	1.9481	3600	0.2082
0.1951	2.0563	3800	0.2058
0.1836	2.1645	4000	0.2048
0.1841	2.2727	4200	0.2043
0.1812	2.3810	4400	0.2028
0.1762	2.4892	4600	0.2025
0.1836	2.5974	4800	0.2017
0.1777	2.7056	5000	0.2012

Framework versions

• PEFT 0.10.0
• Transformers 4.41.0.dev0
• Pytorch 2.2.2+cu118
• Datasets 2.19.0
• Tokenizers 0.19.1