HO_ASR-Model_KIIT2025_v3

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

• Loss: inf
• Wer: 0.5075
• Cer: 0.1778

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: 0.0003
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.7754	0.7526	400	inf	0.7222	0.2516
0.5299	1.5042	800	inf	0.6032	0.2017
0.4254	2.2559	1200	inf	0.5733	0.1903
0.4036	3.0075	1600	inf	0.5588	0.1870
0.3805	3.7601	2000	inf	0.5613	0.1887
0.3593	4.5118	2400	inf	0.5075	0.1777
0.3289	5.2634	2800	inf	0.5155	0.1778
0.3259	6.0151	3200	inf	0.5583	0.1901
0.305	6.7676	3600	inf	0.5089	0.1736

Framework versions

• Transformers 4.55.4
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4