You can deploy this model and start earning money today!

ehyoussef/
xlsr-300m-darija-adapter

Name: xlsr-300m-darija-adapter
Availability: InStock

safetensors

wav2vec2

generated_from_trainer

ehyoussef/xlsr-300m-darija-adapter

finetune

apache-2.0

Views

No views yet

xlsr-300m-darija-adapter

This model is a fine-tuned version of ehyoussef/xlsr-300m-darija-adapter on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.6425
Wer: 0.4359

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: 5e-06
train_batch_size: 4
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 2
total_train_batch_size: 16
total_eval_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 3
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8177	0.28	200	0.6597	0.4460
0.7874	0.57	400	0.6516	0.4428
0.7941	0.85	600	0.6491	0.4419
0.8076	1.14	800	0.6483	0.4402
0.7993	1.42	1000	0.6408	0.4413
0.8036	1.7	1200	0.6458	0.4374
0.8391	1.99	1400	0.6401	0.4381
0.8056	2.27	1600	0.6396	0.4381
0.7581	2.56	1800	0.6425	0.4360
0.7866	2.84	2000	0.6425	0.4359

Framework versions

Transformers 4.39.0
Pytorch 2.8.0+cu126
Datasets 2.18.0
Tokenizers 0.15.2