wav2vec2-large-mms-1b-evenki-colab

This model is a fine-tuned version of facebook/mms-1b-all on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7563
• Wer: 0.7135

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: 0.001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• 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: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
10.0574	0.0308	100	3.4587	1.0
2.6276	0.0615	200	1.7165	0.9800
1.2219	0.0923	300	1.2281	0.9274
1.8242	0.1230	400	1.2651	0.9368
1.1924	0.1538	500	1.1186	0.9025
1.6112	0.1846	600	1.1522	0.9098
1.0251	0.2153	700	1.0574	0.8868
1.5575	0.2461	800	1.1940	0.9190
1.0224	0.2768	900	1.0436	0.8864
1.565	0.3076	1000	1.2350	0.9180
0.9966	0.3384	1100	1.0639	0.8853
1.5941	0.3691	1200	1.2018	0.9266
1.0152	0.3999	1300	1.0436	0.8789
1.5956	0.4306	1400	1.0546	0.8680
0.9592	0.4614	1500	1.0166	0.8753
1.5209	0.4922	1600	0.9918	0.8717
0.9647	0.5229	1700	0.9670	0.8561
1.5689	0.5537	1800	1.0313	0.8598
0.9309	0.5844	1900	0.9857	0.8557
1.4494	0.6152	2000	1.0370	0.9009
0.9489	0.6460	2100	0.9802	0.8449
1.5404	0.6767	2200	1.0277	0.8730
0.93	0.7075	2300	0.9490	0.8214
1.5351	0.7382	2400	1.0075	0.8489
0.86	0.7690	2500	0.9466	0.8278
1.4075	0.7998	2600	0.9908	0.8450
0.8978	0.8305	2700	0.9783	0.8193
1.5049	0.8613	2800	0.9776	0.8263
0.8631	0.8920	2900	0.9383	0.8185
1.4381	0.9228	3000	0.9541	0.8292
0.9529	0.9536	3100	0.9412	0.8297
1.4904	0.9843	3200	0.9731	0.8218
0.9834	1.0151	3300	0.9499	0.8034
1.0134	1.0458	3400	0.9317	0.8129
1.2938	1.0766	3500	0.9105	0.8237
0.9878	1.1074	3600	0.9214	0.8042
1.1626	1.1381	3700	0.9045	0.8081
1.0864	1.1689	3800	0.9101	0.8063
1.21	1.1996	3900	0.9097	0.7937
1.0242	1.2304	4000	0.9051	0.7949
1.1935	1.2612	4100	0.9088	0.8090
0.9784	1.2919	4200	0.9132	0.8008
1.1719	1.3227	4300	0.8861	0.7930
1.0924	1.3534	4400	0.8914	0.8075
1.3006	1.3842	4500	0.8807	0.7852
0.9768	1.4149	4600	0.8859	0.7892
1.1658	1.4457	4700	0.8922	0.7870
0.9708	1.4765	4800	0.8684	0.7819
1.2968	1.5072	4900	0.8680	0.7780
1.0397	1.5380	5000	0.8960	0.7854
1.3094	1.5687	5100	0.8742	0.7815
0.9274	1.5995	5200	0.8732	0.7778
1.2538	1.6303	5300	0.8639	0.7721
1.0089	1.6610	5400	0.8713	0.7862
1.2453	1.6918	5500	0.8620	0.7782
0.9521	1.7225	5600	0.8551	0.7830
1.3161	1.7533	5700	0.8577	0.7748
1.0274	1.7841	5800	0.8573	0.7795
1.2341	1.8148	5900	0.8379	0.7600
0.9222	1.8456	6000	0.8630	0.7726
1.0861	1.8763	6100	0.8489	0.7634
0.9341	1.9071	6200	0.8411	0.7606
1.1717	1.9379	6300	0.8506	0.7747
0.9205	1.9686	6400	0.8626	0.7655
1.1891	1.9994	6500	0.8571	0.7706
0.8629	2.0301	6600	0.8422	0.7666
1.1854	2.0609	6700	0.8483	0.7644
0.8767	2.0917	6800	0.8215	0.7536
1.2317	2.1224	6900	0.8582	0.7751
0.8572	2.1532	7000	0.8308	0.7569
1.2478	2.1839	7100	0.8172	0.7457
0.8012	2.2147	7200	0.8261	0.7624
1.3722	2.2455	7300	0.8543	0.7608
0.8545	2.2762	7400	0.8267	0.7617
1.1439	2.3070	7500	0.8354	0.7565
0.7752	2.3377	7600	0.8200	0.7540
1.3572	2.3685	7700	0.8276	0.7524
0.7083	2.3993	7800	0.8268	0.7502
1.2719	2.4300	7900	0.8276	0.7600
0.7618	2.4608	8000	0.8153	0.7540
1.1886	2.4915	8100	0.8194	0.7528
0.7902	2.5223	8200	0.8111	0.7518
1.2395	2.5531	8300	0.8252	0.7410
0.8243	2.5838	8400	0.8093	0.7439
1.2454	2.6146	8500	0.8115	0.7400
0.8353	2.6453	8600	0.8060	0.7370
1.1679	2.6761	8700	0.8114	0.7539
0.7987	2.7069	8800	0.8039	0.7467
1.1435	2.7376	8900	0.8152	0.7474
0.8146	2.7684	9000	0.7977	0.7476
1.1233	2.7991	9100	0.7973	0.7388
0.7887	2.8299	9200	0.7969	0.7398
1.2149	2.8607	9300	0.8013	0.7511
0.775	2.8914	9400	0.7857	0.7328
1.2072	2.9222	9500	0.7999	0.7357
0.7742	2.9529	9600	0.7901	0.7322
1.2508	2.9837	9700	0.7914	0.7368
0.8621	3.0145	9800	0.7836	0.7402
0.8507	3.0452	9900	0.7790	0.7290
1.0266	3.0760	10000	0.7882	0.7329
0.8631	3.1067	10100	0.7882	0.7281
1.1433	3.1375	10200	0.7823	0.7307
0.8892	3.1683	10300	0.7873	0.7321
1.0211	3.1990	10400	0.7836	0.7273
0.9307	3.2298	10500	0.7818	0.7248
1.1533	3.2605	10600	0.7757	0.7266
0.8969	3.2913	10700	0.7786	0.7198
1.0713	3.3221	10800	0.7787	0.7170
0.8816	3.3528	10900	0.7803	0.7199
1.1048	3.3836	11000	0.7777	0.7199
0.8707	3.4143	11100	0.7750	0.7225
0.9713	3.4451	11200	0.7789	0.7177
0.7791	3.4759	11300	0.7733	0.7170
1.0315	3.5066	11400	0.7691	0.7187
0.7899	3.5374	11500	0.7681	0.7180
0.9405	3.5681	11600	0.7661	0.7175
0.8365	3.5989	11700	0.7629	0.7196
1.1294	3.6297	11800	0.7638	0.7153
0.8537	3.6604	11900	0.7647	0.7114
1.1291	3.6912	12000	0.7593	0.7162
0.7939	3.7219	12100	0.7618	0.7110
1.1158	3.7527	12200	0.7617	0.7128
0.8674	3.7835	12300	0.7583	0.7118
1.0904	3.8142	12400	0.7600	0.7138
0.779	3.8450	12500	0.7574	0.7116
0.9692	3.8757	12600	0.7569	0.7116
0.7636	3.9065	12700	0.7564	0.7138
1.0034	3.9373	12800	0.7577	0.7138
0.7543	3.9680	12900	0.7566	0.7140
1.0205	3.9988	13000	0.7563	0.7135

Framework versions

• Transformers 4.52.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1