xtreme_s_xlsr_t5lephone-small_minds14.en-all

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the GOOGLE/XTREME_S - MINDS14.ALL dataset. It achieves the following results on the evaluation set:

Loss: 0.5979
F1: 0.8918
Accuracy: 0.8921

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: 2
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 8
total_train_batch_size: 32
total_eval_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 1500
num_epochs: 150.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Accuracy
2.3561	2.98	200	2.5464	0.0681	0.1334
1.1851	5.97	400	1.5056	0.5583	0.5861
1.2805	8.95	600	1.1397	0.7106	0.7044
1.0801	11.94	800	0.9863	0.7132	0.7198
0.9285	14.92	1000	0.9912	0.7037	0.7139
0.4164	17.91	1200	0.8226	0.7743	0.7741
0.7669	20.89	1400	0.8131	0.7783	0.7788
0.4606	23.88	1600	0.8314	0.7879	0.7792
0.6975	26.86	1800	0.7667	0.7927	0.7939
0.9913	29.85	2000	0.9207	0.7734	0.7707
0.2307	32.83	2200	0.7651	0.8072	0.8086
0.1412	35.82	2400	0.7132	0.8352	0.8311
0.2141	38.8	2600	0.7551	0.8276	0.8262
0.2169	41.79	2800	0.7900	0.8148	0.8160
0.3942	44.77	3000	0.8621	0.8130	0.8042
0.2306	47.76	3200	0.6788	0.8264	0.8253
0.0975	50.74	3400	0.7236	0.8295	0.8289
0.0062	53.73	3600	0.6872	0.8286	0.8277
0.1781	56.71	3800	0.6990	0.8393	0.8390
0.0309	59.7	4000	0.6348	0.8496	0.8500
0.0026	62.68	4200	0.6737	0.8585	0.8566
0.0043	65.67	4400	0.7780	0.8416	0.8387
0.0032	68.65	4600	0.6899	0.8482	0.8461
0.0302	71.64	4800	0.6813	0.8515	0.8495
0.0027	74.62	5000	0.7163	0.8530	0.8529
0.1165	77.61	5200	0.6249	0.8603	0.8595
0.0021	80.59	5400	0.6747	0.8588	0.8578
0.2558	83.58	5600	0.7514	0.8581	0.8581
0.0162	86.57	5800	0.6782	0.8667	0.8664
0.1929	89.55	6000	0.6371	0.8615	0.8600
0.0621	92.54	6200	0.8079	0.8600	0.8607
0.0017	95.52	6400	0.7072	0.8678	0.8669
0.0008	98.51	6600	0.7323	0.8572	0.8541
0.1655	101.49	6800	0.6953	0.8521	0.8505
0.01	104.48	7000	0.7149	0.8665	0.8674
0.0135	107.46	7200	0.8990	0.8523	0.8488
0.0056	110.45	7400	0.7320	0.8673	0.8664
0.0023	113.43	7600	0.7108	0.8700	0.8705
0.0025	116.42	7800	0.6464	0.8818	0.8820
0.0003	119.4	8000	0.6985	0.8706	0.8713
0.0048	122.39	8200	0.6620	0.8765	0.8740
0.2335	125.37	8400	0.6515	0.8832	0.8828
0.0005	128.36	8600	0.6961	0.8776	0.8762
0.0003	131.34	8800	0.5990	0.8878	0.8882
0.0002	134.33	9000	0.6236	0.8887	0.8889
0.002	137.31	9200	0.6671	0.8847	0.8845
0.0002	140.3	9400	0.5970	0.8931	0.8935
0.0002	143.28	9600	0.6095	0.8906	0.8913
0.0002	146.27	9800	0.6056	0.8910	0.8913
0.0002	149.25	10000	0.5979	0.8918	0.8921

Framework versions

Transformers 4.20.0.dev0
Pytorch 1.11.0+cu113
Datasets 2.1.0
Tokenizers 0.12.1

Splend1dchan/xtreme_s_xlsr_t5lephone-small_minds14.en-all

xtreme_s_xlsr_t5lephone-small_minds14.en-all

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

为此模型做种