helldivers2-jarvis-asrV4

This model is a fine-tuned version of facebook/wav2vec2-base-960h on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 25.6691
• Wer: 0.17
• Cer: 0.7603

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: 1e-05
• train_batch_size: 12
• eval_batch_size: 12
• 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: 50
• num_epochs: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
507.0049	1.0	50	245.1068	0.3743	0.7814
334.0895	2.0	100	161.0169	0.3129	0.7744
279.7883	3.0	150	130.1554	0.2886	0.7716
244.374	4.0	200	104.5179	0.2714	0.7701
211.0537	5.0	250	94.9101	0.2471	0.7683
214.1148	6.0	300	86.9921	0.2514	0.7681
192.2161	7.0	350	75.8522	0.2357	0.7663
173.9121	8.0	400	75.7438	0.2243	0.7658
166.7699	9.0	450	68.1917	0.2243	0.7653
157.0137	10.0	500	59.1111	0.2171	0.7647
166.9989	11.0	550	50.2394	0.2143	0.7643
147.1322	12.0	600	45.1821	0.2171	0.7639
142.869	13.0	650	43.5846	0.2143	0.7641
131.6637	14.0	700	43.0457	0.2071	0.7635
126.8534	15.0	750	45.8238	0.21	0.7635
121.8376	16.0	800	36.9011	0.2029	0.7630
123.6633	17.0	850	38.0945	0.2043	0.7630
104.7618	18.0	900	37.8898	0.2014	0.7626
111.8574	19.0	950	32.1545	0.1986	0.7619
105.2309	20.0	1000	34.3187	0.1943	0.7617
99.6247	21.0	1050	32.9372	0.1843	0.7613
92.9916	22.0	1100	31.1413	0.1871	0.7613
102.2719	23.0	1150	28.5850	0.1814	0.7610
93.6705	24.0	1200	31.5977	0.1814	0.7611
96.9849	25.0	1250	34.8815	0.1843	0.7615
105.9482	26.0	1300	28.3215	0.1757	0.7608
94.6159	27.0	1350	26.6447	0.1771	0.7606
83.7973	28.0	1400	27.4755	0.1814	0.7611
85.906	29.0	1450	23.9488	0.1757	0.7605
87.2979	30.0	1500	23.0084	0.1771	0.7605
77.1649	31.0	1550	22.0342	0.17	0.7603
76.6646	32.0	1600	22.6523	0.1729	0.7604
84.5785	33.0	1650	26.1900	0.1743	0.7604
87.0745	34.0	1700	24.5733	0.1729	0.7603
80.8229	35.0	1750	26.2349	0.1743	0.7605
79.2393	36.0	1800	24.3964	0.1714	0.7603
74.8672	37.0	1850	23.2604	0.1671	0.7602
80.8669	38.0	1900	19.7348	0.17	0.7602
77.9821	39.0	1950	21.5322	0.1686	0.7604
70.6975	40.0	2000	26.5959	0.17	0.7603
73.8925	41.0	2050	24.7372	0.17	0.7602
82.1079	42.0	2100	24.7751	0.1686	0.7602
71.1623	43.0	2150	25.9074	0.1729	0.7604
82.281	44.0	2200	25.5285	0.1714	0.7604
69.5761	45.0	2250	24.7130	0.1714	0.7603
80.4887	46.0	2300	25.2155	0.1686	0.7602
74.3038	47.0	2350	24.2761	0.1671	0.7601
69.2056	48.0	2400	24.0605	0.1671	0.7601
67.871	49.0	2450	25.4337	0.17	0.7602
73.1627	50.0	2500	24.6985	0.1671	0.7601
67.3236	51.0	2550	27.8274	0.1686	0.7603
66.3366	52.0	2600	24.6496	0.17	0.7602
76.2498	53.0	2650	23.5383	0.1714	0.7603
76.629	54.0	2700	24.6941	0.1671	0.7601
64.3516	55.0	2750	24.4289	0.1686	0.7602
64.3968	56.0	2800	24.8174	0.1686	0.7602
73.7818	57.0	2850	26.2579	0.1686	0.7602
74.5849	58.0	2900	26.1643	0.1671	0.7602
65.7819	59.0	2950	25.5072	0.1671	0.7602
73.0206	60.0	3000	25.6691	0.17	0.7603

Framework versions

• Transformers 4.51.3
• Pytorch 2.4.1+cu118
• Datasets 3.5.1
• Tokenizers 0.21.1