ai-light-dance_drums_ft_pretrain_wav2vec2-base-new-v6-1

This model is a fine-tuned version of gary109/ai-light-dance_drums_ft_pretrain_wav2vec2-base-new-v6-1 on the GARY109/AI_LIGHT_DANCE - ONSET-IDMT-MDB-ENST2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5661
• Wer: 0.3416

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: 0.0001
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 100.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2764	1.0	141	0.7125	0.3619
0.5415	2.0	282	0.7252	0.3682
0.3324	3.0	423	0.6779	0.3728
0.4244	4.0	564	0.7403	0.3737
0.5234	5.0	705	0.8086	0.3534
0.3339	6.0	846	0.7187	0.3619
0.5016	7.0	987	0.8582	0.3602
0.3376	8.0	1128	0.8801	0.3674
0.3507	9.0	1269	0.8524	0.3560
0.4844	10.0	1410	0.7152	0.3648
0.4282	11.0	1551	0.6719	0.3475
0.4398	12.0	1692	0.7130	0.3686
0.331	13.0	1833	0.6425	0.3627
0.4488	14.0	1974	0.6483	0.3648
0.3876	15.0	2115	0.6375	0.3509
0.3361	16.0	2256	0.6791	0.3703
0.344	17.0	2397	0.7279	0.3551
0.3198	18.0	2538	0.6801	0.3509
0.2753	19.0	2679	0.6239	0.3509
0.2962	20.0	2820	0.7419	0.3442
0.7503	21.0	2961	0.7279	0.3501
0.4013	22.0	3102	0.6899	0.3792
0.5134	23.0	3243	0.6572	0.3787
0.3144	24.0	3384	0.5882	0.3543
0.3534	25.0	3525	0.5661	0.3416
0.2555	26.0	3666	0.5977	0.3589
0.3524	27.0	3807	0.5953	0.3585
0.314	28.0	3948	0.6359	0.3593
0.2565	29.0	4089	0.6192	0.3615
0.5023	30.0	4230	0.6229	0.3378
0.3025	31.0	4371	0.6002	0.3442
0.3329	32.0	4512	0.6235	0.3513
0.3744	33.0	4653	0.5782	0.3416
0.2899	34.0	4794	0.5835	0.3336
0.306	35.0	4935	0.6061	0.3496
0.2519	36.0	5076	0.5958	0.3652
0.3201	37.0	5217	0.5778	0.3652
0.3011	38.0	5358	0.6238	0.3589
0.2882	39.0	5499	0.6501	0.3361
0.2542	40.0	5640	0.6341	0.3488
0.2717	41.0	5781	0.5890	0.3530
0.3197	42.0	5922	0.5877	0.3471
0.2816	43.0	6063	0.6614	0.3420
0.3301	44.0	6204	0.6334	0.3475
0.2466	45.0	6345	0.6663	0.3429
0.2908	46.0	6486	0.5941	0.3475
0.2785	47.0	6627	0.6337	0.3568
0.2361	48.0	6768	0.5845	0.3399
0.4729	49.0	6909	0.6466	0.3425
0.5103	50.0	7050	0.7112	0.3416
0.2676	51.0	7191	0.6260	0.3307
0.3533	52.0	7332	0.7327	0.3454
0.3308	53.0	7473	0.7150	0.3277
0.2617	54.0	7614	0.6412	0.3391
0.2901	55.0	7755	0.6225	0.3391
0.2847	56.0	7896	0.7385	0.3391
0.2621	57.0	8037	0.7241	0.3496
0.2477	58.0	8178	0.6957	0.3429
0.3147	59.0	8319	0.6808	0.3425
0.3761	60.0	8460	0.6710	0.3450
0.2609	61.0	8601	0.6629	0.3345
0.388	62.0	8742	0.6688	0.3463
0.3684	63.0	8883	0.7018	0.3340
0.2494	64.0	9024	0.6611	0.3399
0.2641	65.0	9165	0.6828	0.3399
0.2716	66.0	9306	0.6409	0.3294
0.2595	67.0	9447	0.6056	0.3231
0.2683	68.0	9588	0.6203	0.3332
0.2571	69.0	9729	0.6484	0.3336
0.2593	70.0	9870	0.6597	0.3294
0.229	71.0	10011	0.6354	0.3235
0.281	72.0	10152	0.6398	0.3294
0.3779	73.0	10293	0.6871	0.3345
0.2998	74.0	10434	0.7329	0.3323
0.2095	75.0	10575	0.7365	0.3239
0.247	76.0	10716	0.6384	0.3290
0.2095	77.0	10857	0.6703	0.3345
0.2074	78.0	10998	0.6577	0.3425
0.2519	79.0	11139	0.6359	0.3370
0.2046	80.0	11280	0.6222	0.3256
1.3195	81.0	11421	0.6126	0.3345
0.2821	82.0	11562	0.6193	0.3294
0.3256	83.0	11703	0.6140	0.3336
0.2743	84.0	11844	0.6204	0.3290
0.2761	85.0	11985	0.6599	0.3252
0.224	86.0	12126	0.6580	0.3294
0.2106	87.0	12267	0.6298	0.3294
0.2706	88.0	12408	0.6411	0.3281
0.2523	89.0	12549	0.6243	0.3264
0.3635	90.0	12690	0.6297	0.3290
0.353	91.0	12831	0.6145	0.3235
0.2491	92.0	12972	0.6296	0.3197
0.1999	93.0	13113	0.6329	0.3222
0.2417	94.0	13254	0.6200	0.3222
0.2397	95.0	13395	0.6137	0.3269
0.2275	96.0	13536	0.6237	0.3277
0.207	97.0	13677	0.6230	0.3235
0.2704	98.0	13818	0.6239	0.3281
0.2119	99.0	13959	0.6224	0.3277
0.2561	100.0	14100	0.6187	0.3269

Framework versions

• Transformers 4.25.0.dev0
• Pytorch 1.8.1+cu111
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2