miugod/mbert_trim_ende_mlm

mbert_trim_ende

该模型基于bert-base-multilingual-cased，使用TextPruner对词表进行裁剪，保留iwslt14德英数据集，用于测试bert-fused的翻译效果。 并且在iwslt14德英数据集上进行掩码语言模型微调，数据的拼接方式是： de, en, de[sep]en, en[sep]de。

Model Details

lang：德英 vocab_size: 119547 -> 21443 model_size: 682M -> 392M iwslt14 de_en BLEU: ?--- tags: - generated_from_trainer metrics: - accuracy model-index: - name: mbert_trim_ende results: []

mbert_trim_ende

This model is a fine-tuned version of miugod/mbert_trim_ende on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8260
• Accuracy: 0.8200

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-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1.0
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.29.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.11.0
• Tokenizers 0.13.3