ssc-led-model

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4926
• Cer: 0.3156
• Wer: 0.7297

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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: 100
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
6.557	0.1390	100	3.2742	0.9964	0.9999
3.4632	0.2780	200	3.2694	0.9912	0.9994
3.4078	0.4170	300	3.2839	0.9842	0.9999
3.3831	0.5559	400	3.1970	0.9833	1.0
3.2858	0.6949	500	3.2411	0.9913	0.9998
3.1984	0.8339	600	2.8873	0.9778	0.9999
3.058	0.9729	700	3.1117	0.9838	1.0
2.9646	1.1112	800	2.7581	0.6400	1.0181
2.9609	1.2502	900	2.7426	0.6333	1.0380
2.9051	1.3892	1000	2.7028	0.6283	1.1000
2.7531	1.5281	1100	2.4181	0.6132	1.0336
2.6876	1.6671	1200	2.4824	0.6167	1.0435
2.5923	1.8061	1300	2.2197	0.5982	0.9760
2.4445	1.9451	1400	2.2790	0.5738	1.0593
2.4327	2.0834	1500	2.0705	0.6024	0.9433
2.2965	2.2224	1600	1.9625	0.5488	1.0015
2.251	2.3614	1700	1.9497	0.5375	0.9985
2.2063	2.5003	1800	1.8847	0.5316	1.0227
2.1696	2.6393	1900	1.8468	0.5320	0.9897
2.0999	2.7783	2000	1.8188	0.5418	0.9231
2.106	2.9173	2100	1.7915	0.5226	0.9906
2.0443	3.0556	2200	1.7832	0.5175	0.9703
2.0008	3.1946	2300	1.9252	0.6242	0.9561
1.9581	3.3336	2400	1.7066	0.5050	0.9349
1.9088	3.4726	2500	1.6904	0.5084	0.9193
1.9802	3.6115	2600	1.7291	0.5031	0.9391
1.9073	3.7505	2700	1.6498	0.4955	0.9059
1.8964	3.8895	2800	1.6772	0.4835	0.9511
1.8309	4.0278	2900	1.5232	0.4640	0.9101
1.8294	4.1668	3000	1.5878	0.4740	0.9068
1.7896	4.3058	3100	1.5236	0.4635	0.8732
1.7279	4.4448	3200	1.5316	0.4659	0.8761
1.8116	4.5837	3300	1.5484	0.4494	0.9161
1.7621	4.7227	3400	1.4950	0.4332	0.8972
1.7334	4.8617	3500	1.4884	0.4439	0.8759
1.8286	5.0	3600	1.5486	0.4640	0.8888
1.68	5.1390	3700	1.5825	0.4565	0.9353
1.6566	5.2780	3800	1.5937	0.4680	0.8777
1.6644	5.4170	3900	1.4593	0.4405	0.8580
1.6664	5.5559	4000	1.5155	0.4474	0.8805
1.6503	5.6949	4100	1.6064	0.4754	0.8777
1.6167	5.8339	4200	1.6223	0.4630	0.9703
1.643	5.9729	4300	1.4182	0.4370	0.8597
1.477	6.1112	4400	1.4183	0.4144	0.8682
1.5618	6.2502	4500	1.3755	0.4162	0.8499
1.5647	6.3892	4600	1.4729	0.4105	0.9161
1.5552	6.5281	4700	1.3810	0.4108	0.8432
1.5455	6.6671	4800	1.3562	0.4252	0.8416
1.5155	6.8061	4900	1.3853	0.4014	0.8696
1.5147	6.9451	5000	1.4001	0.3984	0.8850
1.5683	7.0834	5100	1.2923	0.3984	0.8197
1.5013	7.2224	5200	1.3053	0.3908	0.8516
1.4608	7.3614	5300	1.3336	0.3946	0.8734
1.4684	7.5003	5400	1.2944	0.3925	0.8324
1.4936	7.6393	5500	1.3272	0.3852	0.8324
1.4579	7.7783	5600	1.3462	0.4061	0.8576
1.4586	7.9173	5700	1.3090	0.3859	0.8459
1.3563	8.0556	5800	1.3592	0.4145	0.8451
1.4024	8.1946	5900	1.3865	0.3983	0.9042
1.3459	8.3336	6000	1.4191	0.3923	0.8889
1.3772	8.4726	6100	1.3182	0.3893	0.8290
1.3714	8.6115	6200	1.3219	0.4035	0.8205
1.4096	8.7505	6300	1.3486	0.3844	0.8369
1.3768	8.8895	6400	1.2831	0.3810	0.8239
1.3808	9.0278	6500	1.2448	0.3705	0.7927
1.3053	9.1668	6600	1.3261	0.3886	0.8214
1.3379	9.3058	6700	1.2955	0.3741	0.8301
1.2732	9.4448	6800	1.2895	0.3771	0.8382
1.3168	9.5837	6900	1.2404	0.3669	0.8116
1.311	9.7227	7000	1.2535	0.3774	0.8159
1.3142	9.8617	7100	1.2507	0.3743	0.8225
1.4334	10.0	7200	1.2819	0.3881	0.8157
1.2647	10.1390	7300	1.3880	0.3891	0.8348
1.2282	10.2780	7400	1.2940	0.3763	0.8100
1.2965	10.4170	7500	1.3343	0.3774	0.8548
1.2611	10.5559	7600	1.3572	0.3801	0.8199
1.2283	10.6949	7700	1.3654	0.3708	0.8183
1.2699	10.8339	7800	1.2622	0.3679	0.7898
1.2802	10.9729	7900	1.3185	0.3849	0.8088
1.1242	11.1112	8000	1.2298	0.3617	0.7915
1.1502	11.2502	8100	1.2286	0.3525	0.7668
1.1642	11.3892	8200	1.2514	0.3581	0.8106
1.2153	11.5281	8300	1.2211	0.3544	0.7931
1.1378	11.6671	8400	1.2571	0.3652	0.7997
1.1795	11.8061	8500	1.2248	0.3527	0.7937
1.2308	11.9451	8600	1.2390	0.3582	0.7932
1.238	12.0834	8700	1.2046	0.3513	0.7879
1.0873	12.2224	8800	1.1623	0.3424	0.7721
1.1585	12.3614	8900	1.1803	0.3414	0.7795
1.1275	12.5003	9000	1.2063	0.3443	0.7714
1.1469	12.6393	9100	1.1775	0.3444	0.7778
1.1079	12.7783	9200	1.1507	0.3379	0.7576
1.1546	12.9173	9300	1.1499	0.3429	0.7713
1.064	13.0556	9400	1.2232	0.3445	0.7816
1.0603	13.1946	9500	1.2726	0.3697	0.7919
1.0666	13.3336	9600	1.2574	0.3545	0.8084
1.0512	13.4726	9700	1.2035	0.3532	0.7834
1.0698	13.6115	9800	1.2578	0.3527	0.7963
1.0912	13.7505	9900	1.2416	0.3466	0.7682
1.0627	13.8895	10000	1.2337	0.3453	0.7595
1.0987	14.0278	10100	1.1615	0.3369	0.7522
0.9958	14.1668	10200	1.1879	0.3412	0.7602
0.9942	14.3058	10300	1.1657	0.3421	0.7499
0.9759	14.4448	10400	1.1370	0.3482	0.7603
0.999	14.5837	10500	1.2919	0.3459	0.7803
1.0084	14.7227	10600	1.1895	0.3436	0.7621
1.017	14.8617	10700	1.1151	0.3301	0.7453
1.1085	15.0	10800	1.1994	0.3421	0.7856
0.9106	15.1390	10900	1.2353	0.3411	0.7535
0.9494	15.2780	11000	1.2391	0.3534	0.7680
0.9865	15.4170	11100	1.2126	0.3436	0.7789
0.9331	15.5559	11200	1.2147	0.3458	0.7521
0.9954	15.6949	11300	1.3586	0.3612	0.8058
0.9439	15.8339	11400	1.3371	0.3745	0.7877
0.9967	15.9729	11500	1.3378	0.3844	0.7991
0.8549	16.1112	11600	1.2175	0.3354	0.7517
0.8896	16.2502	11700	1.2278	0.3397	0.7657
0.8664	16.3892	11800	1.2651	0.3352	0.7603
0.8753	16.5281	11900	1.1716	0.3290	0.7444
0.8977	16.6671	12000	1.2236	0.3365	0.7479
0.9496	16.8061	12100	1.1973	0.3331	0.7412
0.91	16.9451	12200	1.1738	0.3343	0.7534
0.9119	17.0834	12300	1.1986	0.3325	0.7440
0.8525	17.2224	12400	1.1816	0.3435	0.7520
0.8101	17.3614	12500	1.1608	0.3279	0.7349
0.8687	17.5003	12600	1.1822	0.3338	0.7440
0.8453	17.6393	12700	1.1843	0.3258	0.7545
0.8405	17.7783	12800	1.1765	0.3392	0.7460
0.8665	17.9173	12900	1.1781	0.3304	0.7403
0.7923	18.0556	13000	1.2630	0.3339	0.7420
0.8042	18.1946	13100	1.2656	0.3343	0.7425
0.7833	18.3336	13200	1.2872	0.3359	0.7512
0.8003	18.4726	13300	1.2731	0.3374	0.7513
0.8029	18.6115	13400	1.2229	0.3304	0.7499
0.8214	18.7505	13500	1.2203	0.3259	0.7501
0.7935	18.8895	13600	1.2336	0.3246	0.7390
0.7993	19.0278	13700	1.2858	0.3251	0.7368
0.716	19.1668	13800	1.2969	0.3221	0.7418
0.7621	19.3058	13900	1.2338	0.3268	0.7489
0.7576	19.4448	14000	1.2330	0.3272	0.7390
0.7413	19.5837	14100	1.1901	0.3265	0.7326
0.7763	19.7227	14200	1.2638	0.3293	0.7475
0.7648	19.8617	14300	1.1992	0.3358	0.7456
0.8236	20.0	14400	1.1934	0.3420	0.7650
0.6941	20.1390	14500	1.2392	0.3316	0.7339
0.6681	20.2780	14600	1.3000	0.3249	0.7423
0.698	20.4170	14700	1.1913	0.3200	0.7260
0.7228	20.5559	14800	1.3172	0.3292	0.7378
0.7112	20.6949	14900	1.2621	0.3281	0.7527
0.7178	20.8339	15000	1.3336	0.3263	0.7634
0.7221	20.9729	15100	1.3298	0.3360	0.7485
0.601	21.1112	15200	1.3518	0.3298	0.7473
0.6322	21.2502	15300	1.3088	0.3367	0.7504
0.6384	21.3892	15400	1.2725	0.3278	0.7371
0.679	21.5281	15500	1.3340	0.3287	0.7537
0.6686	21.6671	15600	1.2990	0.3241	0.7401
0.666	21.8061	15700	1.3288	0.3265	0.7418
0.6466	21.9451	15800	1.2629	0.3161	0.7303
0.7039	22.0834	15900	1.2848	0.3208	0.7347
0.6094	22.2224	16000	1.3526	0.3224	0.7461
0.6243	22.3614	16100	1.3610	0.3253	0.7391
0.6053	22.5003	16200	1.3213	0.3196	0.7347
0.6238	22.6393	16300	1.3298	0.3202	0.7341
0.6083	22.7783	16400	1.3180	0.3180	0.7365
0.5848	22.9173	16500	1.3468	0.3208	0.7457
0.5828	23.0556	16600	1.3385	0.3257	0.7404
0.5904	23.1946	16700	1.3450	0.3191	0.7400
0.6162	23.3336	16800	1.3983	0.3278	0.7377
0.5515	23.4726	16900	1.3701	0.3212	0.7412
0.5598	23.6115	17000	1.3807	0.3215	0.7438
0.5535	23.7505	17100	1.4148	0.3301	0.7419
0.6105	23.8895	17200	1.3604	0.3197	0.7393
0.5633	24.0278	17300	1.3609	0.3193	0.7379
0.5327	24.1668	17400	1.3563	0.3185	0.7348
0.5824	24.3058	17500	1.4036	0.3231	0.7402
0.5035	24.4448	17600	1.3861	0.3218	0.7316
0.5411	24.5837	17700	1.3858	0.3218	0.7436
0.5418	24.7227	17800	1.3647	0.3193	0.7374
0.5533	24.8617	17900	1.4045	0.3211	0.7354
0.5557	25.0	18000	1.3967	0.3248	0.7414
0.493	25.1390	18100	1.4083	0.3184	0.7389
0.4945	25.2780	18200	1.3569	0.3174	0.7350
0.5113	25.4170	18300	1.4117	0.3242	0.7432
0.5373	25.5559	18400	1.4016	0.3206	0.7307
0.5291	25.6949	18500	1.4614	0.3221	0.7430
0.5006	25.8339	18600	1.3930	0.3189	0.7280
0.5286	25.9729	18700	1.4034	0.3225	0.7407
0.4113	26.1112	18800	1.4622	0.3169	0.7278
0.4876	26.2502	18900	1.4270	0.3192	0.7333
0.4888	26.3892	19000	1.4628	0.3157	0.7301
0.4572	26.5281	19100	1.4566	0.3175	0.7319
0.4721	26.6671	19200	1.4436	0.3208	0.7409
0.474	26.8061	19300	1.4377	0.3218	0.7391
0.4796	26.9451	19400	1.4474	0.3208	0.7378
0.5193	27.0834	19500	1.4400	0.3168	0.7309
0.4976	27.2224	19600	1.4349	0.3159	0.7302
0.4634	27.3614	19700	1.4753	0.3174	0.7339
0.4465	27.5003	19800	1.4669	0.3168	0.7345
0.4511	27.6393	19900	1.4651	0.3209	0.7390
0.4475	27.7783	20000	1.4643	0.3182	0.7328
0.4309	27.9173	20100	1.4598	0.3161	0.7303
0.4391	28.0556	20200	1.4789	0.3152	0.7373
0.452	28.1946	20300	1.4677	0.3145	0.7280
0.4376	28.3336	20400	1.4662	0.3179	0.7307
0.446	28.4726	20500	1.4773	0.3156	0.7318
0.4332	28.6115	20600	1.4617	0.3166	0.7297
0.4197	28.7505	20700	1.4659	0.3175	0.7336
0.4086	28.8895	20800	1.4816	0.3177	0.7359
0.4456	29.0278	20900	1.4862	0.3176	0.7349
0.4198	29.1668	21000	1.5056	0.3153	0.7317
0.4435	29.3058	21100	1.4887	0.3155	0.7289
0.3888	29.4448	21200	1.4954	0.3151	0.7261
0.4202	29.5837	21300	1.4947	0.3154	0.7289
0.4353	29.7227	21400	1.4926	0.3161	0.7300
0.4154	29.8617	21500	1.4921	0.3161	0.7295
0.4489	30.0	21600	1.4926	0.3156	0.7297

Framework versions

• Transformers 4.57.2
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.22.0