XLS-R-300m-SV

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_7_0 - SV-SE dataset. It achieves the following results on the evaluation set:

• Loss: 0.3171
• Wer: 0.2468

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: 7.5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• num_epochs: 50.0
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.16.0.dev0
• Pytorch 1.10.0+cu102
• Datasets 1.17.1.dev0
• Tokenizers 0.10.3

Evaluation Commands

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

python eval.py --model_id hf-test/xls-r-300m-sv --dataset mozilla-foundation/common_voice_7_0 --config sv-SE --split test

2. To evaluate on speech-recognition-community-v2/dev_data

python eval.py --model_id hf-test/xls-r-300m-sv --dataset speech-recognition-community-v2/dev_data --config sv --split validation --chunk_length_s 5.0 --stride_length_s 1.0

Inference With LM

import torch
from datasets import load_dataset
from transformers import AutoModelForCTC, AutoProcessor
import torchaudio.functional as F


model_id = "hf-test/xls-r-300m-sv"

sample_iter = iter(load_dataset("mozilla-foundation/common_voice_7_0", "sv-SE", 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
# => "jag lämnade grovjobbet åt honom"

Eval results on Common Voice 7 "test" (WER):