wav2vec2-large-xlsr-300-arabic

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

• Loss: 0.4514
• Wer: 0.4256
• Cer: 0.1528

Evaluation Commands

1. To evaluate on mozilla-foundation/common_voice_7_0 with split test

python eval.py --model_id kingabzpro/wav2vec2-large-xlsr-300-arabic --dataset mozilla-foundation/common_voice_7_0 --config ur --split test

Inference With LM

import torch
from datasets import load_dataset
from transformers import AutoModelForCTC, AutoProcessor
import torchaudio.functional as F
model_id = "kingabzpro/wav2vec2-large-xlsr-300-arabic"
sample_iter = iter(load_dataset("mozilla-foundation/common_voice_8_0", "ar", split="test", streaming=True, use_auth_token=True))
sample = next(sample_iter)
resampled_audio = F.resample(torch.tensor(sample["audio"]["array"]), 48_000, 16_000).numpy()
model = AutoModelForCTC.from_pretrained(model_id)
processor = AutoProcessor.from_pretrained(model_id)
input_values = processor(resampled_audio, return_tensors="pt").input_values
with torch.no_grad():
    logits = model(input_values).logits
transcription = processor.batch_decode(logits.numpy()).text

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0003
• train_batch_size: 64
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
5.4375	1.8	500	3.3330	1.0	1.0
2.2187	3.6	1000	0.7790	0.6501	0.2338
0.9471	5.4	1500	0.5353	0.5015	0.1822
0.7416	7.19	2000	0.4889	0.4490	0.1640
0.6358	8.99	2500	0.4514	0.4256	0.1528

Framework versions

• Transformers 4.17.0.dev0
• Pytorch 1.10.2+cu102
• Datasets 1.18.2.dev0
• Tokenizers 0.11.0