ASR-cv-corpus-ug-11

This model is a fine-tuned version of piyazon/ASR-cv-corpus-ug-10 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0206
• Wer: 0.0144

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: 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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0233	0.1668	300	0.0246	0.0255
0.0292	0.3336	600	0.0329	0.0422
0.0304	0.5004	900	0.0294	0.0355
0.0303	0.6672	1200	0.0303	0.0368
0.0338	0.8340	1500	0.0301	0.0346
0.0296	1.0006	1800	0.0270	0.0361
0.0194	1.1674	2100	0.0299	0.0406
0.0208	1.3342	2400	0.0268	0.0313
0.0227	1.5010	2700	0.0295	0.0362
0.0226	1.6678	3000	0.0259	0.0305
0.0216	1.8346	3300	0.0258	0.0309
0.0206	2.0011	3600	0.0263	0.0325
0.0132	2.1679	3900	0.0230	0.0256
0.0157	2.3347	4200	0.0266	0.0313
0.0145	2.5015	4500	0.0242	0.0265
0.015	2.6683	4800	0.0244	0.0287
0.015	2.8351	5100	0.0283	0.0334
0.0154	3.0017	5400	0.0252	0.0296
0.0119	3.1685	5700	0.0230	0.0265
0.0102	3.3353	6000	0.0223	0.0246
0.0094	3.5021	6300	0.0238	0.0239
0.01	3.6689	6600	0.0273	0.0252
0.0104	3.8357	6900	0.0236	0.0226
0.0118	4.0022	7200	0.0212	0.0236
0.0084	4.1690	7500	0.0222	0.0216
0.0074	4.3358	7800	0.0218	0.0217
0.0066	4.5026	8100	0.0210	0.0203
0.0078	4.6694	8400	0.0203	0.0222
0.0073	4.8363	8700	0.0214	0.0225
0.0062	5.0028	9000	0.0193	0.0197
0.0054	5.1696	9300	0.0205	0.0205
0.0052	5.3364	9600	0.0215	0.0210
0.0058	5.5032	9900	0.0210	0.0204
0.0057	5.6700	10200	0.0211	0.0199
0.0045	5.8368	10500	0.0203	0.0193
0.0042	6.0033	10800	0.0197	0.0201
0.0028	6.1701	11100	0.0196	0.0186
0.0038	6.3369	11400	0.0203	0.0186
0.0034	6.5038	11700	0.0193	0.0178
0.0031	6.6706	12000	0.0204	0.0178
0.0033	6.8374	12300	0.0201	0.0181
0.003	7.0039	12600	0.0206	0.0195
0.0027	7.1707	12900	0.0206	0.0173
0.0025	7.3375	13200	0.0199	0.0170
0.0024	7.5043	13500	0.0181	0.0166
0.0016	7.6711	13800	0.0199	0.0160
0.0022	7.8379	14100	0.0191	0.0164
0.0019	8.0044	14400	0.0198	0.0165
0.0016	8.1713	14700	0.0201	0.0162
0.0012	8.3381	15000	0.0197	0.0156
0.0016	8.5049	15300	0.0191	0.0158
0.0014	8.6717	15600	0.0195	0.0156
0.001	8.8385	15900	0.0198	0.0156
0.0008	9.0050	16200	0.0202	0.0148
0.0005	9.1718	16500	0.0211	0.0147
0.0006	9.3386	16800	0.0205	0.0146
0.0006	9.5054	17100	0.0208	0.0142
0.0006	9.6722	17400	0.0205	0.0145
0.0004	9.8390	17700	0.0206	0.0144

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu128
• Datasets 4.1.0
• Tokenizers 0.22.0