metadata

tags:
  - generated_from_trainer
datasets:
  - ai_light_dance
metrics:
  - wer
model-index:
  - name: ai-light-dance_drums_ft_pretrain_wav2vec2-base-new_onset-idmt-2
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: ai_light_dance
          type: ai_light_dance
          config: onset-idmt-2
          split: train
          args: onset-idmt-2
        metrics:
          - name: Wer
            type: wer
            value: 0.26

ai-light-dance_drums_ft_pretrain_wav2vec2-base-new_onset-idmt-2

This model is a fine-tuned version of gary109/ai-light-dance_drums_pretrain_wav2vec2-base on the ai_light_dance dataset. It achieves the following results on the evaluation set:

Loss: 0.5174
Wer: 0.26

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: 0.0003
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 30
num_epochs: 100.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	9	97.9319	1.0
17.1836	2.0	18	45.7229	1.0
13.2869	3.0	27	2.7579	1.0
2.6495	4.0	36	2.7427	1.0
1.7135	5.0	45	2.5477	1.0
1.4609	6.0	54	1.7126	1.0
1.374	7.0	63	1.3668	0.9967
1.2951	8.0	72	1.1274	0.9867
1.0493	9.0	81	0.7346	0.5178
0.8835	10.0	90	0.7664	0.4122
0.8835	11.0	99	0.5438	0.3867
0.7019	12.0	108	0.4876	0.3711
0.6906	13.0	117	0.5194	0.36
0.6535	14.0	126	0.4489	0.3556
0.6225	15.0	135	0.4383	0.3333
0.547	16.0	144	0.4521	0.3556
0.5525	17.0	153	0.5476	0.3344
0.6152	18.0	162	0.4466	0.36
0.5055	19.0	171	0.3981	0.3256
0.5204	20.0	180	0.4924	0.3078
0.5204	21.0	189	0.4085	0.32
0.4742	22.0	198	0.4255	0.3233
0.4774	23.0	207	0.4321	0.2889
0.5029	24.0	216	0.4412	0.3167
0.4889	25.0	225	0.4051	0.3044
0.4446	26.0	234	0.3918	0.3089
0.4255	27.0	243	0.4039	0.2956
0.4396	28.0	252	0.4113	0.2956
0.4265	29.0	261	0.5576	0.3022
0.4289	30.0	270	0.3558	0.3078
0.4289	31.0	279	0.3390	0.3167
0.3817	32.0	288	0.3739	0.3422
0.4192	33.0	297	0.3179	0.3056
0.3719	34.0	306	0.3622	0.3033
0.3685	35.0	315	0.4057	0.3256
0.3752	36.0	324	0.3950	0.31
0.378	37.0	333	0.3907	0.3567
0.4438	38.0	342	0.3376	0.31
0.3978	39.0	351	0.3395	0.2833
0.3639	40.0	360	0.3646	0.2856
0.3639	41.0	369	0.3546	0.3044
0.3535	42.0	378	0.3699	0.2889
0.3311	43.0	387	0.3882	0.3022
0.3475	44.0	396	0.4749	0.2889
0.4048	45.0	405	0.3437	0.2911
0.2984	46.0	414	0.3664	0.27
0.3535	47.0	423	0.3291	0.2889
0.3015	48.0	432	0.3538	0.2767
0.3628	49.0	441	0.4411	0.2733
0.3303	50.0	450	0.3425	0.29
0.3303	51.0	459	0.3162	0.3011
0.271	52.0	468	0.3685	0.2933
0.3299	53.0	477	0.4216	0.2933
0.2782	54.0	486	0.4713	0.3044
0.348	55.0	495	0.4310	0.3078
0.2969	56.0	504	0.4898	0.2767
0.2757	57.0	513	0.5195	0.2789
0.2662	58.0	522	0.4631	0.2911
0.2706	59.0	531	0.4275	0.2833
0.2684	60.0	540	0.5535	0.2789
0.2684	61.0	549	0.4733	0.2978
0.2819	62.0	558	0.4969	0.2833
0.2819	63.0	567	0.6202	0.2789
0.2889	64.0	576	0.3955	0.2733
0.2515	65.0	585	0.3806	0.2656
0.2468	66.0	594	0.3473	0.2722
0.2557	67.0	603	0.4170	0.2722
0.2477	68.0	612	0.4749	0.2678
0.2965	69.0	621	0.4387	0.2611
0.2606	70.0	630	0.4586	0.2656
0.2606	71.0	639	0.5755	0.2733
0.2442	72.0	648	0.5582	0.2656
0.347	73.0	657	0.3897	0.2711
0.2444	74.0	666	0.3369	0.2533
0.2811	75.0	675	0.3487	0.2578
0.24	76.0	684	0.3692	0.2589
0.2466	77.0	693	0.4567	0.2578
0.2769	78.0	702	0.4041	0.2633
0.2464	79.0	711	0.3813	0.2622
0.2791	80.0	720	0.3990	0.2556
0.2791	81.0	729	0.3997	0.2489
0.2365	82.0	738	0.4537	0.2533
0.2693	83.0	747	0.5943	0.2611
0.2285	84.0	756	0.5805	0.2656
0.2468	85.0	765	0.5609	0.2656
0.2226	86.0	774	0.5948	0.2667
0.2419	87.0	783	0.5910	0.2544
0.2254	88.0	792	0.5741	0.26
0.2083	89.0	801	0.4984	0.2611
0.2318	90.0	810	0.5093	0.26
0.2318	91.0	819	0.5284	0.2633
0.2458	92.0	828	0.4885	0.2656
0.2394	93.0	837	0.4818	0.2622
0.2018	94.0	846	0.5037	0.26
0.235	95.0	855	0.5011	0.2578
0.2252	96.0	864	0.4931	0.2611
0.2147	97.0	873	0.4881	0.2589
0.2227	98.0	882	0.4956	0.2589
0.2168	99.0	891	0.5097	0.2589
0.2282	100.0	900	0.5174	0.26

Framework versions

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