README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: assis
    results: []

assis

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3836
• Wer: 1

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• 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: 3000
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
23.2159	0.6	100	22.1148	1
18.1848	1.2	200	16.7223	1
9.7817	1.8	300	7.9404	1
4.5091	2.4	400	3.7900	1
3.4946	2.99	500	3.2953	1
3.3286	3.59	600	3.1827	1
3.2078	4.19	700	3.1068	1
3.1528	4.79	800	3.0573	1
3.0709	5.39	900	3.0196	1
3.0163	5.99	1000	2.9919	1
2.9789	6.59	1100	2.9504	1
2.9468	7.19	1200	2.9272	1
2.9389	7.78	1300	2.9129	1
2.9192	8.38	1400	2.9005	1
2.9069	8.98	1500	2.8861	1
2.9074	9.58	1600	2.8816	1
2.883	10.18	1700	2.8746	1
2.8746	10.78	1800	2.8718	1
2.8637	11.38	1900	2.8567	1
2.8613	11.98	2000	2.8570	1
2.8598	12.57	2100	2.8449	1
2.8357	13.17	2200	2.8393	1
2.8352	13.77	2300	2.8350	1
2.8178	14.37	2400	2.7879	1
2.5089	14.97	2500	2.3686	1
2.0826	15.57	2600	1.8915	1
1.6003	16.17	2700	1.3513	1
1.2925	16.77	2800	1.0568	1
1.0837	17.37	2900	0.8760	1
0.9333	17.96	3000	0.7588	1
0.8214	18.56	3100	0.6841	1
0.7302	19.16	3200	0.6099	1
0.6815	19.76	3300	0.5459	1
0.6548	20.36	3400	0.5087	1
0.569	20.96	3500	0.4853	1
0.5919	21.56	3600	0.4666	1
0.5306	22.16	3700	0.4508	1
0.5228	22.75	3800	0.4389	1
0.5263	23.35	3900	0.4287	1
0.4945	23.95	4000	0.4182	1
0.4809	24.55	4100	0.4122	1
0.4813	25.15	4200	0.4112	1
0.4664	25.75	4300	0.3972	1
0.455	26.35	4400	0.3950	1
0.4415	26.95	4500	0.3962	1
0.4399	27.54	4600	0.3930	1
0.4451	28.14	4700	0.3864	1
0.4343	28.74	4800	0.3867	1
0.4418	29.34	4900	0.3865	1
0.4223	29.94	5000	0.3836	1

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3