README.md

metadata

license: apache-2.0
base_model: facebook/wav2vec2-large-xlsr-53
tags:
  - automatic-speech-recognition
  - ./sample_speech.py
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: en-xlsr
    results: []

en-xlsr

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the ./SAMPLE_SPEECH.PY - NA dataset. It achieves the following results on the evaluation set:

• Loss: 0.4835
• Cer: 0.1119
• Wer: 0.2446

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
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• 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: 10
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
2.9534	0.22	100	2.9533	1.0	1.0
2.933	0.44	200	2.9231	1.0	1.0
2.904	0.65	300	2.8851	1.0	1.0
2.3607	0.87	400	2.1546	0.6799	0.9976
1.1725	1.09	500	0.9899	0.2665	0.6191
0.9865	1.31	600	0.8060	0.2126	0.5064
0.8959	1.53	700	0.7131	0.1980	0.4607
0.7743	1.74	800	0.6663	0.1799	0.4370
0.7805	1.96	900	0.6159	0.1683	0.3997
0.6562	2.18	1000	0.6186	0.1537	0.3705
0.6223	2.4	1100	0.5698	0.1496	0.3552
0.5627	2.62	1200	0.5555	0.1446	0.3372
0.5476	2.84	1300	0.5435	0.1416	0.3307
0.5002	3.05	1400	0.5304	0.1436	0.3393
0.5174	3.27	1500	0.5377	0.1485	0.3357
0.4745	3.49	1600	0.5289	0.1340	0.3132
0.5239	3.71	1700	0.5112	0.1395	0.3239
0.5115	3.93	1800	0.5079	0.1342	0.3094
0.4471	4.14	1900	0.5131	0.1301	0.2965
0.4455	4.36	2000	0.5015	0.1278	0.2931
0.4199	4.58	2100	0.4954	0.1299	0.2962
0.4699	4.8	2200	0.4827	0.1268	0.2890
0.3521	5.02	2300	0.4857	0.1217	0.2782
0.3976	5.23	2400	0.4936	0.1231	0.2802
0.365	5.45	2500	0.4906	0.1221	0.2774
0.3857	5.67	2600	0.4843	0.1202	0.2757
0.3578	5.89	2700	0.4857	0.1196	0.2708
0.3298	6.11	2800	0.4867	0.1197	0.2689
0.3099	6.32	2900	0.4924	0.1237	0.2770
0.3606	6.54	3000	0.4851	0.1189	0.2684
0.3807	6.76	3100	0.4700	0.1196	0.2656
0.3286	6.98	3200	0.4770	0.1205	0.2730
0.3318	7.2	3300	0.4845	0.1166	0.2579
0.2936	7.42	3400	0.4909	0.1159	0.2570
0.3119	7.63	3500	0.4899	0.1150	0.2539
0.3142	7.85	3600	0.4782	0.1143	0.2550
0.2935	8.07	3700	0.4885	0.1153	0.2527
0.2805	8.29	3800	0.4906	0.1143	0.2529
0.254	8.51	3900	0.4822	0.1144	0.2538
0.2855	8.72	4000	0.4852	0.1123	0.2476
0.2661	8.94	4100	0.4847	0.1132	0.2496
0.2524	9.16	4200	0.4900	0.1116	0.2442
0.253	9.38	4300	0.4888	0.1120	0.2458
0.2591	9.6	4400	0.4813	0.1125	0.2458
0.2583	9.81	4500	0.4844	0.1114	0.2435

Framework versions

• Transformers 4.34.0
• Pytorch 2.1.0+cu121
• Datasets 2.14.5
• Tokenizers 0.14.1