wav2vec2-large-xlsr-facebook-300m-texts-exp-1

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: 0.7400
• Wer: 0.5455
• Cer: 0.2311
• Per: 0.5289

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Per
38.9028	1.0	14	18.5048	1.0	0.9671	1.0
38.9028	2.0	28	9.8810	1.0	0.9671	1.0
38.9028	3.0	42	7.7650	1.0	0.9671	1.0
38.9028	4.0	56	6.9159	1.0	0.9671	1.0
38.9028	5.0	70	6.3930	1.0	0.9671	1.0
38.9028	6.0	84	5.9372	1.0	0.9671	1.0
38.9028	7.0	98	5.5319	1.0	0.9671	1.0
12.9708	8.0	112	5.1820	1.0	0.9671	1.0
12.9708	9.0	126	4.8767	1.0	0.9671	1.0
12.9708	10.0	140	4.6011	1.0	0.9671	1.0
12.9708	11.0	154	4.3649	1.0	0.9671	1.0
12.9708	12.0	168	4.1568	1.0	0.9671	1.0
12.9708	13.0	182	3.9743	1.0	0.9671	1.0
12.9708	14.0	196	3.8195	1.0	0.9671	1.0
4.98	15.0	210	3.6889	1.0	0.9671	1.0
4.98	16.0	224	3.5786	1.0	0.9671	1.0
4.98	17.0	238	3.4828	1.0	0.9671	1.0
4.98	18.0	252	3.4061	1.0	0.9671	1.0
4.98	19.0	266	3.3420	1.0	0.9671	1.0
4.98	20.0	280	3.2863	1.0	0.9671	1.0
4.98	21.0	294	3.2379	1.0	0.9671	1.0
3.5425	22.0	308	3.1723	1.0	0.9671	1.0
3.5425	23.0	322	3.1202	1.0	0.9671	1.0
3.5425	24.0	336	3.0854	1.0	0.9671	1.0
3.5425	25.0	350	3.0373	1.0	0.9671	1.0
3.5425	26.0	364	3.0513	1.0	0.9671	1.0
3.5425	27.0	378	3.0163	1.0	0.9671	1.0
3.5425	28.0	392	2.9870	1.0	0.9671	1.0
3.0988	29.0	406	2.9741	1.0	0.9671	1.0
3.0988	30.0	420	2.9698	1.0	0.9671	1.0
3.0988	31.0	434	2.9717	1.0	0.9671	1.0
3.0988	32.0	448	2.9575	1.0	0.9671	1.0
3.0988	33.0	462	2.9537	1.0	0.9671	1.0
3.0988	34.0	476	2.9581	1.0	0.9671	1.0
3.0988	35.0	490	2.9491	1.0	0.9671	1.0
2.9753	36.0	504	2.9414	1.0	0.9671	1.0
2.9753	37.0	518	2.9396	1.0	0.9671	1.0
2.9753	38.0	532	2.9386	1.0	0.9671	1.0
2.9753	39.0	546	2.9313	1.0	0.9671	1.0
2.9753	40.0	560	2.9585	1.0	0.9671	1.0
2.9753	41.0	574	2.9350	1.0	0.9671	1.0
2.9753	42.0	588	2.9248	1.0	0.9671	1.0
2.9468	43.0	602	2.9306	1.0	0.9671	1.0
2.9468	44.0	616	2.9244	1.0	0.9671	1.0
2.9468	45.0	630	2.9234	1.0	0.9671	1.0
2.9468	46.0	644	2.9226	1.0	0.9671	1.0
2.9468	47.0	658	2.9173	1.0	0.9671	1.0
2.9468	48.0	672	2.9177	1.0	0.9671	1.0
2.9468	49.0	686	2.9207	1.0	0.9671	1.0
2.9258	50.0	700	2.9186	1.0	0.9671	1.0
2.9258	51.0	714	2.9123	1.0	0.9671	1.0
2.9258	52.0	728	2.9160	1.0	0.9671	1.0
2.9258	53.0	742	2.9112	1.0	0.9671	1.0
2.9258	54.0	756	2.8988	1.0	0.9671	1.0
2.9258	55.0	770	2.9045	1.0	0.9671	1.0
2.9258	56.0	784	2.9132	1.0	0.9671	1.0
2.9258	57.0	798	2.8828	1.0	0.9671	1.0
2.9114	58.0	812	2.8750	1.0	0.9671	1.0
2.9114	59.0	826	2.8658	1.0	0.9671	1.0
2.9114	60.0	840	2.8470	1.0	0.9671	1.0
2.9114	61.0	854	2.8315	1.0	0.9671	1.0
2.9114	62.0	868	2.8104	1.0	0.9671	1.0
2.9114	63.0	882	2.7643	1.0	0.9671	1.0
2.9114	64.0	896	2.7412	1.0	0.9671	1.0
2.8525	65.0	910	2.7178	1.0	0.9671	1.0
2.8525	66.0	924	2.6887	1.0	0.9668	1.0
2.8525	67.0	938	2.6267	1.0	0.9670	1.0
2.8525	68.0	952	2.5971	0.9995	0.9575	0.9995
2.8525	69.0	966	2.5567	0.9997	0.9243	0.9997
2.8525	70.0	980	2.5002	1.0	0.8266	1.0
2.8525	71.0	994	2.4290	1.0	0.7585	1.0
2.6546	72.0	1008	2.3754	1.0	0.7096	1.0
2.6546	73.0	1022	2.3152	1.0	0.6939	1.0
2.6546	74.0	1036	2.2623	1.0	0.6643	1.0
2.6546	75.0	1050	2.1567	1.0	0.6488	1.0
2.6546	76.0	1064	2.0693	1.0	0.6107	1.0
2.6546	77.0	1078	1.9740	1.0	0.5911	1.0
2.6546	78.0	1092	1.9123	0.9997	0.5685	0.9997
2.2421	79.0	1106	1.8264	1.0	0.5500	1.0
2.2421	80.0	1120	1.7678	0.9997	0.5253	0.9997
2.2421	81.0	1134	1.6755	1.0	0.5180	1.0
2.2421	82.0	1148	1.6509	1.0	0.4988	1.0
2.2421	83.0	1162	1.5905	1.0	0.4927	1.0
2.2421	84.0	1176	1.5812	1.0	0.4782	1.0
2.2421	85.0	1190	1.4685	1.0	0.4804	1.0
1.7467	86.0	1204	1.4286	1.0	0.4693	1.0
1.7467	87.0	1218	1.4161	0.9995	0.4555	0.9995
1.7467	88.0	1232	1.3703	1.0	0.4503	1.0
1.7467	89.0	1246	1.3651	0.9980	0.4382	0.9980
1.7467	90.0	1260	1.3306	0.9970	0.4315	0.9967
1.7467	91.0	1274	1.2882	0.9957	0.4274	0.9954
1.7467	92.0	1288	1.2564	0.9939	0.4210	0.9937
1.4044	93.0	1302	1.2472	0.9726	0.3998	0.9695
1.4044	94.0	1316	1.2310	0.9518	0.3813	0.9464
1.4044	95.0	1330	1.1658	0.9563	0.3799	0.9528
1.4044	96.0	1344	1.1555	0.9317	0.3697	0.9264
1.4044	97.0	1358	1.1608	0.8918	0.3573	0.8865
1.4044	98.0	1372	1.1363	0.8720	0.3483	0.8652
1.4044	99.0	1386	1.1085	0.8479	0.3379	0.8405
1.191	100.0	1400	1.0848	0.8489	0.3348	0.8426
1.191	101.0	1414	1.0706	0.8179	0.3252	0.8106
1.191	102.0	1428	1.0839	0.7905	0.3212	0.7824
1.191	103.0	1442	1.0419	0.8187	0.3210	0.8121
1.191	104.0	1456	1.0630	0.7692	0.3141	0.7593
1.191	105.0	1470	1.0117	0.7831	0.3122	0.7743
1.191	106.0	1484	1.0035	0.7633	0.3064	0.7529
1.191	107.0	1498	1.0035	0.7463	0.3017	0.7364
1.0047	108.0	1512	0.9954	0.7395	0.2974	0.7288
1.0047	109.0	1526	0.9830	0.7438	0.2979	0.7349
1.0047	110.0	1540	0.9699	0.7219	0.2917	0.7115
1.0047	111.0	1554	0.9626	0.7184	0.2901	0.7075
1.0047	112.0	1568	0.9367	0.7176	0.2896	0.7075
1.0047	113.0	1582	0.9550	0.7057	0.2838	0.6960
1.0047	114.0	1596	0.9140	0.6872	0.2769	0.6790
0.9031	115.0	1610	0.9214	0.6851	0.2781	0.6752
0.9031	116.0	1624	0.9040	0.6877	0.2786	0.6767
0.9031	117.0	1638	0.9107	0.6920	0.2811	0.6813
0.9031	118.0	1652	0.8947	0.6699	0.2743	0.6585
0.9031	119.0	1666	0.8975	0.6595	0.2715	0.6480
0.9031	120.0	1680	0.8817	0.6750	0.2729	0.6646
0.9031	121.0	1694	0.8686	0.6529	0.2670	0.6412
0.8165	122.0	1708	0.8622	0.6442	0.2653	0.6323
0.8165	123.0	1722	0.8741	0.6437	0.2653	0.6320
0.8165	124.0	1736	0.8656	0.6399	0.2636	0.6282
0.8165	125.0	1750	0.8582	0.6366	0.2628	0.6249
0.8165	126.0	1764	0.8547	0.6366	0.2624	0.6249
0.8165	127.0	1778	0.8515	0.6275	0.2606	0.6158
0.8165	128.0	1792	0.8515	0.6272	0.2603	0.6153
0.7588	129.0	1806	0.8351	0.6204	0.2565	0.6089
0.7588	130.0	1820	0.8311	0.6216	0.2573	0.6089
0.7588	131.0	1834	0.8403	0.6188	0.2579	0.6061
0.7588	132.0	1848	0.8281	0.6153	0.2562	0.6036
0.7588	133.0	1862	0.8210	0.6130	0.2560	0.6008
0.7588	134.0	1876	0.8312	0.6107	0.2566	0.5990
0.7588	135.0	1890	0.8171	0.6122	0.2547	0.6013
0.723	136.0	1904	0.8264	0.6056	0.2551	0.5940
0.723	137.0	1918	0.8117	0.6084	0.2540	0.5957
0.723	138.0	1932	0.8066	0.6018	0.2511	0.5886
0.723	139.0	1946	0.8022	0.6023	0.2525	0.5899
0.723	140.0	1960	0.7986	0.6034	0.2518	0.5912
0.723	141.0	1974	0.7937	0.5945	0.2486	0.5820
0.723	142.0	1988	0.8014	0.5950	0.2504	0.5828
0.6458	143.0	2002	0.7933	0.5907	0.2492	0.5759
0.6458	144.0	2016	0.7918	0.5922	0.2478	0.5777
0.6458	145.0	2030	0.7814	0.5876	0.2456	0.5734
0.6458	146.0	2044	0.7851	0.5866	0.2455	0.5716
0.6458	147.0	2058	0.7908	0.5884	0.2462	0.5741
0.6458	148.0	2072	0.7968	0.5894	0.2475	0.5749
0.6458	149.0	2086	0.7879	0.5851	0.2458	0.5698
0.6326	150.0	2100	0.7997	0.5835	0.2466	0.5681
0.6326	151.0	2114	0.7859	0.5841	0.2454	0.5683
0.6326	152.0	2128	0.7836	0.5785	0.2431	0.5620
0.6326	153.0	2142	0.7778	0.5769	0.2408	0.5617
0.6326	154.0	2156	0.7653	0.5731	0.2394	0.5574
0.6326	155.0	2170	0.7690	0.5790	0.2407	0.5627
0.6326	156.0	2184	0.7688	0.5731	0.2389	0.5574
0.6326	157.0	2198	0.7631	0.5703	0.2383	0.5546
0.583	158.0	2212	0.7584	0.5744	0.2396	0.5587
0.583	159.0	2226	0.7607	0.5691	0.2385	0.5531
0.583	160.0	2240	0.7651	0.5675	0.2381	0.5521
0.583	161.0	2254	0.7572	0.5625	0.2372	0.5467
0.583	162.0	2268	0.7561	0.5637	0.2377	0.5475
0.583	163.0	2282	0.7570	0.5602	0.2354	0.5434
0.583	164.0	2296	0.7545	0.5571	0.2341	0.5411
0.5626	165.0	2310	0.7548	0.5582	0.2350	0.5416
0.5626	166.0	2324	0.7436	0.5561	0.2324	0.5391
0.5626	167.0	2338	0.7545	0.5546	0.2335	0.5378
0.5626	168.0	2352	0.7469	0.5584	0.2342	0.5416
0.5626	169.0	2366	0.7503	0.5536	0.2333	0.5371
0.5626	170.0	2380	0.7522	0.5472	0.2325	0.5305
0.5626	171.0	2394	0.7434	0.5508	0.2321	0.5348
0.5491	172.0	2408	0.7462	0.5495	0.2328	0.5328
0.5491	173.0	2422	0.7524	0.5482	0.2322	0.5317
0.5491	174.0	2436	0.7400	0.5455	0.2311	0.5289
0.5491	175.0	2450	0.7447	0.5510	0.2334	0.5333
0.5491	176.0	2464	0.7500	0.5500	0.2328	0.5320
0.5491	177.0	2478	0.7471	0.5498	0.2331	0.5322
0.5491	178.0	2492	0.7496	0.5513	0.2328	0.5340
0.5339	179.0	2506	0.7521	0.5505	0.2322	0.5330
0.5339	180.0	2520	0.7461	0.5467	0.2315	0.5289
0.5339	181.0	2534	0.7466	0.5460	0.2312	0.5289
0.5339	182.0	2548	0.7436	0.5470	0.2312	0.5305
0.5339	183.0	2562	0.7439	0.5427	0.2308	0.5254
0.5339	184.0	2576	0.7428	0.5434	0.2309	0.5267
0.5339	185.0	2590	0.7456	0.5439	0.2309	0.5267
0.5208	186.0	2604	0.7444	0.5449	0.2308	0.5272
0.5208	187.0	2618	0.7438	0.5444	0.2306	0.5269
0.5208	188.0	2632	0.7487	0.5439	0.2310	0.5272
0.5208	189.0	2646	0.7471	0.5422	0.2300	0.5254
0.5208	190.0	2660	0.7436	0.5422	0.2304	0.5249
0.5208	191.0	2674	0.7428	0.5414	0.2299	0.5236
0.5208	192.0	2688	0.7454	0.5414	0.2295	0.5239
0.5432	193.0	2702	0.7460	0.5442	0.2299	0.5267
0.5432	194.0	2716	0.7450	0.5409	0.2295	0.5231

Framework versions

• Transformers 4.28.0
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.13.3