learning_rate: 0.0001
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
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_steps: 5000
num_epochs: 30
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.35	0.5831	1000	3.2692	100.0	99.4209
3.142	1.1662	2000	3.1542	100.0	99.4212
1.4813	1.7493	3000	1.0182	91.5214	31.4839
1.1133	2.3324	4000	0.7708	81.7330	23.7124
0.894	2.9155	5000	0.6496	76.0178	20.7264
0.9352	3.4985	6000	0.6090	74.9243	20.1963
0.8816	4.0816	7000	0.5506	69.7262	17.9071
0.801	4.6647	8000	0.5120	66.1408	17.0636
0.7132	5.2478	9000	0.4882	64.1662	16.2953
0.654	5.8309	10000	0.4576	61.4582	15.5443
0.7286	6.4140	11000	0.4507	61.0758	15.3310
0.7422	6.9971	12000	0.4585	61.7722	15.2274
0.7389	7.5802	13000	0.5055	65.4444	16.0938
0.7023	8.1633	14000	0.4984	64.6668	15.9446
0.7115	8.7464	15000	0.4921	64.1089	15.8478
0.7052	9.3294	16000	0.4869	63.5880	15.7628
0.7082	9.9125	17000	0.4822	63.0782	15.6848
0.8213	10.4956	18000	0.4779	62.5942	15.6096
0.7041	11.0787	19000	0.4737	62.2469	15.5443
0.7276	11.6618	20000	0.4676	61.8018	15.4780
0.6596	12.2449	21000	0.4619	61.4988	15.4447
0.7221	12.8280	22000	0.4572	61.4138	15.4201
0.7638	13.4111	23000	0.4514	61.2531	15.3809
0.7648	13.9942	24000	0.4474	60.9742	15.2795
0.6832	14.5773	25000	0.4444	60.7489	15.2099
0.6201	15.1603	26000	0.4425	60.5697	15.1334
0.6104	15.7434	27000	0.4406	60.4533	15.0807
0.7071	16.3265	28000	0.4397	60.3979	15.0542
0.6756	16.9096	29000	0.4385	60.1910	15.0046
0.7406	17.4927	30000	0.4370	60.0358	14.9430
0.6657	18.0758	31000	0.4357	59.9989	14.9082
0.6175	18.6589	32000	0.4354	59.9509	14.9023
0.7055	19.2420	33000	0.4340	59.8807	14.8761
0.7032	19.8251	34000	0.4342	59.8881	14.8699
0.7579	20.4082	35000	0.4342	59.8382	14.8632
0.7546	20.9913	36000	0.4312	59.5186	14.7775
0.6248	21.5743	37000	0.4298	59.4835	14.7313
0.656	22.1574	38000	0.4292	59.5316	14.7716
0.6904	22.7405	39000	0.4283	59.4318	14.7294
0.7032	23.3236	40000	0.4287	59.3875	14.7143
0.7106	23.9067	41000	0.4279	59.4022	14.7211
0.627	24.4898	42000	0.4278	59.3468	14.6986
0.6056	25.0729	43000	0.4280	59.3838	14.7112
0.6176	25.6560	44000	0.4278	59.3505	14.6943
0.689	26.2391	45000	0.4278	59.3616	14.6927
0.6782	26.8222	46000	0.4281	59.3598	14.6940
0.7306	27.4052	47000	0.4278	59.3265	14.6801
0.6847	27.9883	48000	0.4278	59.3505	14.6878
0.619	28.5714	49000	0.4278	59.3394	14.6857
0.6997	29.1545	50000	0.4278	59.3339	14.6816
0.678	29.7376	51000	0.4278	59.3099	14.6801

Framework versions

Transformers 4.48.1
Pytorch 2.5.1+cu121
Datasets 3.2.0
Tokenizers 0.21.0

davidilag/wav2vec2-xls-r-300m-scandinavian-E2-100h-30-epochs-20250129_v1.1

wav2vec2-xls-r-300m-scandinavian-E2-100h-30-epochs-20250129_v1.1

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

davidilag/
wav2vec2-xls-r-300m-scandinavian-E2-100h-30-epochs-20250129_v1.1