ssc-ruc-xlsr300m-model-mix-adapt-max-lowlr

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.0445
• Cer: 0.1910
• Wer: 0.7159

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: 2
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• 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: 100
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.25	0.5944	200	0.8255	0.2102	0.7596
0.217	1.1872	400	0.8573	0.2098	0.7456
0.1829	1.7816	600	0.8578	0.2071	0.7451
0.1507	2.3744	800	0.8909	0.2013	0.7319
0.1311	2.9688	1000	0.9180	0.1977	0.7385
0.0985	3.5617	1200	0.9911	0.1988	0.7200
0.083	4.1545	1400	1.0063	0.1898	0.7127
0.0642	4.7489	1600	1.0445	0.1910	0.7159

Framework versions

• Transformers 4.57.2
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.22.0