wav2vec2-large-xls-r-kurmanji_aug-telephone_v5

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2454
• Wer: 0.6290

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: 8e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• 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: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
7.815	0.1443	200	5.5186	1.0
3.3411	0.2887	400	3.3207	1.0
3.1552	0.4330	600	3.0841	1.0
3.0904	0.5773	800	3.0660	1.0
3.0635	0.7216	1000	3.0024	1.0
2.7642	0.8660	1200	2.5705	0.9995
1.9992	1.0101	1400	1.9583	0.9515
1.8396	1.1544	1600	1.7276	0.8935
1.765	1.2988	1800	1.6737	0.8551
1.7256	1.4431	2000	1.5843	0.8350
1.6413	1.5874	2200	1.5684	0.8216
1.5743	1.7317	2400	1.5523	0.8149
1.5836	1.8761	2600	1.4723	0.8035
1.4436	2.0202	2800	1.5032	0.7954
1.5497	2.1645	3000	1.4220	0.7815
1.5002	2.3089	3200	1.4112	0.7741
1.5095	2.4532	3400	1.4244	0.7685
1.5096	2.5975	3600	1.3701	0.7705
1.4866	2.7418	3800	1.3748	0.7646
1.3959	2.8862	4000	1.4268	0.7681
1.5534	3.0303	4200	1.3444	0.7613
1.4359	3.1746	4400	1.3694	0.7424
1.3374	3.3190	4600	1.3692	0.7405
1.3404	3.4633	4800	1.3744	0.7427
1.4218	3.6076	5000	1.3273	0.7338
1.3039	3.7519	5200	1.3755	0.7331
1.3425	3.8963	5400	1.3124	0.7279
1.4135	4.0404	5600	1.3247	0.7223
1.3104	4.1847	5800	1.3698	0.7221
1.3563	4.3291	6000	1.3019	0.7176
1.2269	4.4734	6200	1.3040	0.7135
1.2845	4.6177	6400	1.3283	0.7189
1.2872	4.7620	6600	1.3023	0.7143
1.4382	4.9064	6800	1.3155	0.7165
1.388	5.0505	7000	1.2720	0.7055
1.3685	5.1948	7200	1.2953	0.7068
1.2421	5.3392	7400	1.2655	0.7068
1.2857	5.4835	7600	1.2946	0.6995
1.2746	5.6278	7800	1.2790	0.6980
1.2232	5.7721	8000	1.2647	0.6957
1.2888	5.9165	8200	1.2688	0.6923
1.264	6.0606	8400	1.2880	0.6947
1.2393	6.2049	8600	1.2652	0.6920
1.2347	6.3493	8800	1.3297	0.6957
1.2623	6.4936	9000	1.2662	0.6880
1.2888	6.6379	9200	1.2479	0.6955
1.2617	6.7822	9400	1.2539	0.6905
1.2417	6.9266	9600	1.2786	0.6814
1.303	7.0707	9800	1.2602	0.6845
1.1906	7.2150	10000	1.2565	0.6824
1.1864	7.3594	10200	1.2617	0.6849
1.1854	7.5037	10400	1.2731	0.6812
1.1525	7.6480	10600	1.2272	0.6922
1.2142	7.7924	10800	1.2500	0.6798
1.247	7.9367	11000	1.2556	0.6816
1.1305	8.0808	11200	1.2819	0.6838
1.1322	8.2251	11400	1.3057	0.6858
1.1797	8.3695	11600	1.2393	0.6769
1.1337	8.5138	11800	1.2620	0.6782
1.2728	8.6581	12000	1.2422	0.6727
1.1648	8.8025	12200	1.2427	0.6766
1.2013	8.9468	12400	1.2361	0.6730
1.1283	9.0909	12600	1.2718	0.6801
1.197	9.2353	12800	1.2518	0.6751
1.1701	9.3796	13000	1.2475	0.6676
1.1722	9.5239	13200	1.2492	0.6680
1.1057	9.6682	13400	1.2268	0.6693
1.2214	9.8126	13600	1.2230	0.6636
1.1606	9.9569	13800	1.2294	0.6668
1.2363	10.1010	14000	1.2762	0.6740
1.1314	10.2454	14200	1.2585	0.6670
1.1285	10.3897	14400	1.2381	0.6595
1.1105	10.5340	14600	1.2270	0.6648
1.1566	10.6783	14800	1.2133	0.6629
1.1405	10.8227	15000	1.2194	0.6627
1.1321	10.9670	15200	1.2274	0.6629
1.1513	11.1111	15400	1.2163	0.6587
1.0144	11.2555	15600	1.2394	0.6596
1.1714	11.3998	15800	1.2306	0.6534
1.0912	11.5441	16000	1.2176	0.6513
1.1228	11.6884	16200	1.2280	0.6512
1.0333	11.8328	16400	1.2177	0.6529
1.1128	11.9771	16600	1.2129	0.6567
1.081	12.1212	16800	1.2443	0.6549
1.0753	12.2656	17000	1.2444	0.6553
1.1234	12.4099	17200	1.2570	0.6532
1.0857	12.5542	17400	1.2448	0.6579
1.0638	12.6985	17600	1.2422	0.6503
1.058	12.8429	17800	1.2605	0.6552
1.0955	12.9872	18000	1.2095	0.6517
1.0803	13.1313	18200	1.2585	0.6548
1.0747	13.2757	18400	1.2272	0.6509
1.0812	13.4200	18600	1.2323	0.6506
1.0908	13.5643	18800	1.2551	0.6528
1.0275	13.7086	19000	1.2790	0.6586
1.1142	13.8530	19200	1.2268	0.6523
1.0585	13.9973	19400	1.2435	0.6516
1.028	14.1414	19600	1.2391	0.6499
1.0052	14.2858	19800	1.2482	0.6503
1.0836	14.4301	20000	1.2608	0.6502
1.0855	14.5744	20200	1.2717	0.6540
0.9843	14.7187	20400	1.2418	0.6480
1.0577	14.8631	20600	1.2249	0.6415
0.9924	15.0072	20800	1.2510	0.6509
1.0052	15.1515	21000	1.2525	0.6489
1.0604	15.2959	21200	1.2139	0.6456
1.0508	15.4402	21400	1.2373	0.6426
1.0659	15.5845	21600	1.2365	0.6459
0.9982	15.7288	21800	1.2425	0.6456
1.0367	15.8732	22000	1.2218	0.6434
0.947	16.0173	22200	1.2406	0.6456
0.9553	16.1616	22400	1.2621	0.6491
1.001	16.3060	22600	1.2358	0.6443
1.0512	16.4503	22800	1.2422	0.6409
1.0767	16.5946	23000	1.2405	0.6421
0.9993	16.7390	23200	1.2561	0.6465
1.1164	16.8833	23400	1.2498	0.6441
0.9514	17.0274	23600	1.2390	0.6412
0.995	17.1717	23800	1.2395	0.6402
1.0164	17.3161	24000	1.2607	0.6443
1.0391	17.4604	24200	1.2437	0.6406
1.057	17.6047	24400	1.2593	0.6442
0.9618	17.7491	24600	1.2319	0.6388
1.0072	17.8934	24800	1.2342	0.6394
0.9628	18.0375	25000	1.2617	0.6431
0.9822	18.1819	25200	1.2329	0.6353
1.0308	18.3262	25400	1.2278	0.6359
0.924	18.4705	25600	1.2377	0.6393
0.8818	18.6148	25800	1.2282	0.6344
0.9586	18.7592	26000	1.2327	0.6377
1.0566	18.9035	26200	1.2240	0.6372
1.0085	19.0476	26400	1.2467	0.6354
0.9729	19.1920	26600	1.2339	0.6347
0.9759	19.3363	26800	1.2319	0.6351
0.9842	19.4806	27000	1.2330	0.6350
0.9716	19.6249	27200	1.2423	0.6357
0.9777	19.7693	27400	1.2291	0.6360
0.9884	19.9136	27600	1.2322	0.6353
0.9929	20.0577	27800	1.2399	0.6332
0.9958	20.2021	28000	1.2410	0.6327
1.0149	20.3464	28200	1.2415	0.6325
0.993	20.4907	28400	1.2359	0.6328
0.9304	20.6350	28600	1.2479	0.6358
1.0002	20.7794	28800	1.2400	0.6334
0.9608	20.9237	29000	1.2343	0.6332
0.9917	21.0678	29200	1.2538	0.6346
1.0764	21.2122	29400	1.2507	0.6342
0.9517	21.3565	29600	1.2522	0.6324
1.0005	21.5008	29800	1.2452	0.6313
0.9288	21.6451	30000	1.2450	0.6289
0.9411	21.7895	30200	1.2445	0.6303
0.9789	21.9338	30400	1.2426	0.6299
0.9787	22.0779	30600	1.2387	0.6299
0.9344	22.2223	30800	1.2563	0.6321
0.9543	22.3666	31000	1.2456	0.6291
0.9442	22.5109	31200	1.2471	0.6301
0.93	22.6552	31400	1.2437	0.6304
0.9141	22.7996	31600	1.2323	0.6289
0.9566	22.9439	31800	1.2540	0.6322
0.9897	23.0880	32000	1.2508	0.6307
0.9457	23.2324	32200	1.2563	0.6297
0.9621	23.3767	32400	1.2455	0.6291
0.933	23.5210	32600	1.2506	0.6291
1.0252	23.6653	32800	1.2407	0.6282
0.9206	23.8097	33000	1.2530	0.6278
1.0066	23.9540	33200	1.2440	0.6279
0.9313	24.0981	33400	1.2611	0.6298
0.8982	24.2425	33600	1.2491	0.6299
0.9105	24.3868	33800	1.2464	0.6292
0.9753	24.5311	34000	1.2395	0.6281
0.9289	24.6754	34200	1.2465	0.6297
0.8978	24.8198	34400	1.2421	0.6288
0.9517	24.9641	34600	1.2454	0.6290

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2