helldivers2-jarvis-asrV2

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: 28.5574
• Wer: 0.2202
• Cer: 0.8428

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
564.7336	1.0	50	298.1647	0.4697	0.8612
417.1263	2.0	100	214.3165	0.3927	0.8549
361.7804	3.0	150	187.2314	0.3596	0.8524
326.2224	4.0	200	145.8555	0.3450	0.8509
288.2027	5.0	250	126.9637	0.3413	0.8501
297.8872	6.0	300	99.9533	0.3174	0.8486
260.7991	7.0	350	91.3213	0.3046	0.8480
248.5329	8.0	400	89.1852	0.2936	0.8480
228.2494	9.0	450	71.1274	0.2881	0.8471
235.5672	10.0	500	74.1389	0.2661	0.8465
231.3615	11.0	550	64.1308	0.2697	0.8461
214.3394	12.0	600	63.4379	0.2587	0.8456
216.4062	13.0	650	65.4323	0.2385	0.8453
207.2749	14.0	700	51.0200	0.2385	0.8446
194.7204	15.0	750	53.9227	0.2495	0.8449
191.7318	16.0	800	46.7860	0.2404	0.8446
184.4219	17.0	850	47.1186	0.2459	0.8443
174.2516	18.0	900	50.4025	0.2385	0.8443
181.3694	19.0	950	51.3427	0.2367	0.8440
171.3787	20.0	1000	57.2478	0.2349	0.8444
169.6002	21.0	1050	45.4265	0.2367	0.8440
163.5564	22.0	1100	57.1685	0.2294	0.8441
179.7925	23.0	1150	42.7982	0.2220	0.8437
160.0045	24.0	1200	43.9563	0.2275	0.8436
162.1235	25.0	1250	41.3504	0.2349	0.8437
171.0586	26.0	1300	35.8618	0.2294	0.8433
163.222	27.0	1350	48.1241	0.2275	0.8436
144.6168	28.0	1400	35.0105	0.2239	0.8435
154.0386	29.0	1450	40.7426	0.2312	0.8437
149.5638	30.0	1500	37.7159	0.2440	0.8438
152.7088	31.0	1550	44.9629	0.2202	0.8429
141.1782	32.0	1600	43.5452	0.2202	0.8431
148.6998	33.0	1650	46.8319	0.2257	0.8433
156.1795	34.0	1700	40.1366	0.2239	0.8432
134.192	35.0	1750	48.7881	0.2275	0.8433
136.9826	36.0	1800	50.8378	0.2202	0.8431
132.9241	37.0	1850	28.7557	0.2183	0.8425
141.7361	38.0	1900	33.2380	0.2220	0.8429
133.5196	39.0	1950	42.5577	0.2239	0.8429
131.6621	40.0	2000	33.2488	0.2275	0.8429
132.694	41.0	2050	32.1173	0.2239	0.8428
136.4332	42.0	2100	31.2864	0.2183	0.8426
138.5151	43.0	2150	43.6833	0.2220	0.8427
133.53	44.0	2200	27.9468	0.2183	0.8424
119.6547	45.0	2250	43.3999	0.2147	0.8426
134.2982	46.0	2300	28.5882	0.2202	0.8428
129.6781	47.0	2350	40.8014	0.2165	0.8426
133.2878	48.0	2400	46.5926	0.2183	0.8425
120.2284	49.0	2450	30.5833	0.2183	0.8426
131.5662	50.0	2500	40.5421	0.2202	0.8430
128.9309	51.0	2550	33.1733	0.2202	0.8426
125.6526	52.0	2600	33.8879	0.2220	0.8429
134.5112	53.0	2650	31.5242	0.2183	0.8425
128.9252	54.0	2700	36.8484	0.2239	0.8430
120.8643	55.0	2750	35.2391	0.2183	0.8426
124.6056	56.0	2800	41.8901	0.2183	0.8424
128.6048	57.0	2850	34.9353	0.2257	0.8427
137.4	58.0	2900	36.6512	0.2183	0.8427
112.7822	59.0	2950	37.5492	0.2220	0.8426
132.3333	60.0	3000	28.5574	0.2202	0.8428

Framework versions

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