Saad-Wazir24/indic-slid-mhubert-task2

indic-slid-mhubert

This model is a fine-tuned version of utter-project/mHuBERT-147 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.8465
• Accuracy: 0.6139

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: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 16
• 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: 7

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
49.4727	0.3682	100	3.0906	0.0473
49.3094	0.7365	200	3.0860	0.0903
47.9158	1.1031	300	2.9682	0.22
45.1127	1.4713	400	2.8256	0.2988
42.5860	1.8396	500	2.7005	0.3612
41.5817	2.2062	600	2.5785	0.4191
40.0492	2.5745	700	2.4995	0.4279
37.1908	2.9427	800	2.3784	0.4967
33.6228	3.3093	900	2.2773	0.5282
33.5105	3.6776	1000	2.2024	0.5394
31.8522	4.0442	1100	2.1260	0.5421
31.5263	4.4124	1200	2.0781	0.55
26.6307	4.7807	1300	2.0186	0.5812
26.1449	5.1473	1400	1.9504	0.5827
24.5869	5.5155	1500	1.9050	0.6003
29.2614	5.8838	1600	1.8901	0.6067
25.2821	6.2504	1700	1.8729	0.6085
26.8669	6.6186	1800	1.8508	0.6079
24.7780	6.9869	1900	1.8465	0.6139

Framework versions

• Transformers 5.2.0
• Pytorch 2.9.0+cu126
• Datasets 2.21.0
• Tokenizers 0.22.2