whisper-large-v3-turbo-gurmukhi-lora

This model is a fine-tuned version of openai/whisper-large-v3-turbo on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6912

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_BNB 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
• training_steps: 1000
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss
19.2302	0.4069	50	2.3483
16.1382	0.8138	100	2.0084
15.0314	1.2197	150	1.8724
14.5433	1.6267	200	1.8118
13.9557	2.0326	250	1.7802
14.0038	2.4395	300	1.7612
13.9163	2.8464	350	1.7460
13.7943	3.2523	400	1.7359
13.6503	3.6592	450	1.7269
13.5450	4.0651	500	1.7190
13.4929	4.4720	550	1.7140
13.4390	4.8789	600	1.7090
13.5675	5.2848	650	1.7042
13.4880	5.6918	700	1.7008
13.2943	6.0977	750	1.6979
13.5647	6.5046	800	1.6958
13.3725	6.9115	850	1.6939
13.3143	7.3174	900	1.6925
13.4461	7.7243	950	1.6915
13.3344	8.1302	1000	1.6912

Framework versions

• PEFT 0.18.1
• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.5
• Tokenizers 0.22.2