learning_rate: 2e-05
train_batch_size: 20
eval_batch_size: 16
seed: 3407
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.02
num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Ter	Wer
1.0002	0.1860	400	0.9297	11846.1538	120.0
0.3177	0.3721	800	0.3096	11384.6154	113.8462
0.2278	0.5581	1200	0.2437	11230.7692	112.3077
0.1923	0.7442	1600	0.1919	10769.2308	107.6923
0.1674	0.9302	2000	0.1675	9846.1538	98.4615
0.1313	1.1163	2400	0.1594	10923.0769	109.2308
0.128	1.3023	2800	0.1370	10615.3846	106.1538
0.1241	1.4884	3200	0.1289	9692.3077	96.9231
0.1177	1.6744	3600	0.1167	9538.4615	95.3846
0.1092	1.8605	4000	0.1082	10615.3846	106.1538
0.0817	2.0465	4400	0.1035	9538.4615	95.3846
0.0773	2.2326	4800	0.0986	10000.0	100.0
0.0769	2.4186	5200	0.0957	8923.0769	89.2308
0.0783	2.6047	5600	0.0894	10461.5385	104.6154
0.0712	2.7907	6000	0.0906	8923.0769	89.2308
0.0655	2.9767	6400	0.0840	8769.2308	87.6923
0.0588	3.1628	6800	0.0771	7076.9231	70.7692
0.0588	3.3488	7200	0.0646	7538.4615	75.3846
0.0517	3.5349	7600	0.0647	8000.0	80.0
0.0541	3.7209	8000	0.0645	8461.5385	84.6154
0.0536	3.9070	8400	0.0679	8461.5385	84.6154
0.0414	4.0930	8800	0.0691	8923.0769	89.2308
0.0431	4.2791	9200	0.0609	8769.2308	87.6923
0.0404	4.4651	9600	0.0607	8153.8462	81.5385
0.0395	4.6512	10000	0.0631	8615.3846	86.1538
0.038	4.8372	10400	0.0594	8615.3846	86.1538
0.0308	5.0233	10800	0.0577	8769.2308	87.6923
0.0312	5.2093	11200	0.0542	8153.8462	81.5385
0.0306	5.3953	11600	0.0543	8769.2308	87.6923
0.029	5.5814	12000	0.0525	8307.6923	83.0769
0.0299	5.7674	12400	0.0531	8307.6923	83.0769
0.0311	5.9535	12800	0.0487	8307.6923	83.0769
0.0218	6.1395	13200	0.0475	7846.1538	78.4615
0.0245	6.3256	13600	0.0500	8923.0769	89.2308
0.0256	6.5116	14000	0.0500	8769.2308	87.6923
0.0219	6.6977	14400	0.0497	80.0	8000.0
0.0214	6.8837	14800	0.0502	81.5385	8153.8462
0.021	7.0698	15200	0.0498	81.5385	8153.8462
0.0215	7.2558	15600	0.0494	81.5385	8153.8462
0.0215	7.4419	16000	0.0493	78.4615	7846.1538
0.0202	7.6279	16400	0.0499	81.5385	8153.8462
0.0202	7.8140	16800	0.0496	81.5385	8153.8462
0.0216	8.0	17200	0.0498	84.6154	8461.5385

Framework versions

Transformers 4.57.1
Pytorch 2.8.0+cu128
Datasets 3.6.0
Tokenizers 0.22.1

Tnaot/bopha-stt-1.3.1

bopha-stt-1.3.1

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Tnaot/
bopha-stt-1.3.1