Edit model card

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
Downloads last month
23
Hosted inference API
or or
This model can be loaded on the Inference API on-demand.

Dataset used to train kingabzpro/wav2vec2-large-xlsr-300-arabic

Evaluation results