Indonesian NusaBERT, BERT, BART, RoBERTa, GPT2
Collection
BERT, BART, RoBERTa, GPT2 models pre-trained/fine-tuned on Indonesian datasets • 15 items • Updated
Configuration Parsing Warning:Config file config.json cannot be fetched (too big)
Configuration Parsing Warning:Config file tokenizer_config.json cannot be fetched (too big)
NusaBert-v1.3.1
NusaBert is a pretrained ModernBERT architecture from scratch using 80B English-Indonesian datasets. It also uses a tokenizer pretrained on this bilingual dataset. Thew model has a maximum context length of 8192 (512 in the last 100k steps) and achieves the following results on the evaluation set:
- Loss: 1.5935
- Accuracy: 0.6678
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: 64
- eval_batch_size: 64
- seed: 4
- distributed_type: multi-GPU
- num_devices: 8
- total_train_batch_size: 512
- total_eval_batch_size: 512
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- training_steps: 100000
Training results
Framework versions
- Transformers 4.48.3
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.0
- Downloads last month
- 25
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for cahya/NusaBert-v1.3.1
Collection including cahya/NusaBert-v1.3.1
Evaluation results
- Accuracy on cahya/id-en-mediumself-reported0.668