job_108_wav2vec-20250720_032152-sw

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

• Loss: 229.3007
• Score Wer: 1.0
• Score Cer: 1.0

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: 1
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Score Wer	Score Cer
906.4417	1.0	100	2881.3474	1.0299	1.2024
895.8637	2.0	200	2822.2612	1.3704	1.4005
647.5116	3.0	300	1568.8075	1.0	1.0
317.8601	4.0	400	669.0593	1.0	1.0
239.066	5.0	500	504.7881	1.0	1.0
192.0934	6.0	600	415.7474	1.0	1.0
150.1755	7.0	700	338.2761	1.0	1.0
121.2479	8.0	800	284.7086	1.0	1.0
97.5205	9.0	900	258.2477	1.0	1.0
91.999	10.0	1000	247.1204	1.0	1.0
88.8228	11.0	1100	246.0690	1.0	1.0
82.9214	12.0	1200	241.6842	1.0	1.0
94.9128	13.0	1300	229.5241	1.0	1.0
82.7699	14.0	1400	229.2999	1.0	1.0
90.9523	15.0	1500	229.2943	1.0	1.0
85.804	16.0	1600	229.2999	1.0	1.0
84.2813	17.0	1700	229.3109	1.0	1.0
89.636	18.0	1800	229.2934	1.0	1.0
86.5602	19.0	1900	229.3026	1.0	1.0
88.333	20.0	2000	229.3133	1.0	1.0
84.5754	21.0	2100	229.3123	1.0	1.0
95.6529	22.0	2200	229.3007	1.0	1.0

Framework versions

• Transformers 4.53.2
• Pytorch 2.7.1+cu126
• Datasets 4.0.0
• Tokenizers 0.21.2