learning_rate: 1e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 32
optimizer: Use adamw_torch 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
0.707	0.1208	200	0.5108	183.9537
0.1483	0.2415	400	0.2337	78.3214
0.0999	0.3623	600	0.1841	66.4900
0.079	0.4830	800	0.1643	61.2645
0.0672	0.6038	1000	0.1451	57.8383
0.0622	0.7245	1200	0.1369	54.7942
0.0592	0.8453	1400	0.1288	53.8575
0.0551	0.9660	1600	0.1234	52.0951
0.0407	1.0863	1800	0.1252	51.5652
0.0386	1.2071	2000	0.1237	55.1393
0.0375	1.3278	2200	0.1201	49.9877
0.0382	1.4486	2400	0.1215	50.6902
0.0351	1.5694	2600	0.1178	50.1849
0.0347	1.6901	2800	0.1172	48.0035
0.0333	1.8109	3000	0.1133	48.3362
0.0322	1.9316	3200	0.1112	47.2270
0.0218	2.0519	3400	0.1145	47.3626
0.0204	2.1727	3600	0.1166	45.7358
0.0194	2.2934	3800	0.1200	47.0791
0.0196	2.4142	4000	0.1198	47.1161
0.019	2.5349	4200	0.1198	46.5738
0.0191	2.6557	4400	0.1196	47.5721
0.0202	2.7765	4600	0.1210	47.9788
0.0183	2.8972	4800	0.1210	48.6443
0.016	3.0175	5000	0.1225	46.0069
0.0104	3.1383	5200	0.1274	46.4876
0.0107	3.2590	5400	0.1316	45.5632
0.0108	3.3798	5600	0.1314	48.7306
0.011	3.5005	5800	0.1338	46.1301
0.011	3.6213	6000	0.1311	46.5985
0.0104	3.7420	6200	0.1379	46.3766
0.0106	3.8628	6400	0.1377	46.3397
0.0117	3.9835	6600	0.1337	46.1671
0.0056	4.1038	6800	0.1476	45.6618
0.0056	4.2246	7000	0.1503	48.4841
0.0055	4.3454	7200	0.1495	46.7094
0.006	4.4661	7400	0.1492	45.9330
0.0057	4.5869	7600	0.1522	46.2041
0.0062	4.7076	7800	0.1543	46.7833
0.0062	4.8284	8000	0.1537	46.0316
0.0064	4.9491	8200	0.1578	47.9418
0.0028	5.0694	8400	0.1646	48.1760
0.0027	5.1902	8600	0.1667	46.0562
0.0031	5.3109	8800	0.1665	46.2287
0.0032	5.4317	9000	0.1690	45.8960
0.0033	5.5525	9200	0.1679	45.4770
0.0033	5.6732	9400	0.1713	45.7604
0.0029	5.7940	9600	0.1710	45.9946
0.0037	5.9147	9800	0.1716	46.0562
0.0023	6.0350	10000	0.1778	45.9330
0.0016	6.1558	10200	0.1818	46.8203
0.0016	6.2765	10400	0.1876	46.8450
0.0014	6.3973	10600	0.1860	45.6125
0.0015	6.5180	10800	0.1886	47.4612
0.0015	6.6388	11000	0.1899	46.7957
0.0016	6.7595	11200	0.1896	45.6495
0.0015	6.8803	11400	0.1897	44.9717
0.0015	7.0006	11600	0.1907	48.2253
0.0006	7.1214	11800	0.1999	47.6214
0.0008	7.2421	12000	0.2007	45.4400
0.0006	7.3629	12200	0.2064	46.0562
0.0007	7.4836	12400	0.2002	48.0404
0.0006	7.6044	12600	0.2060	45.6248
0.0007	7.7251	12800	0.2035	45.2428
0.0009	7.8459	13000	0.2045	48.1144
0.0007	7.9666	13200	0.2050	45.2798
0.0004	8.0869	13400	0.2073	47.3256
0.0003	8.2077	13600	0.2110	47.3010
0.0004	8.3285	13800	0.2167	45.0333
0.0004	8.4492	14000	0.2141	45.2798
0.0003	8.5700	14200	0.2160	45.4030
0.0003	8.6907	14400	0.2171	45.2181
0.0003	8.8115	14600	0.2173	44.6882
0.0003	8.9322	14800	0.2173	47.7077
0.0001	9.0525	15000	0.2216	46.6601
0.0001	9.1733	15200	0.2238	45.2058
0.0001	9.2940	15400	0.2258	45.5016
0.0001	9.4148	15600	0.2257	45.5509
0.0001	9.5355	15800	0.2253	45.6248
0.0001	9.6563	16000	0.2263	45.2428
0.0001	9.7771	16200	0.2256	45.3660
0.0001	9.8978	16400	0.2260	44.9840

Framework versions

Transformers 4.50.3
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1

kavyamanohar/whisper-small-ml-codemixed

whisper-small-ml-codemixed

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

kavyamanohar/
whisper-small-ml-codemixed