BLOOM | French
Model description
This is a test model Large Language Model using architecture from BLOOM and dataset from bigscience-data/roots_fr_wikipedia.
Intended uses & limitations
the model can only serve French language.
Training and evaluation data
Dataset:
- bigscience-data/roots_fr_wikipedia
Training Hardware:
- 2x Tesla T4 GPU from Kaggle
Eval:
- eval_loss: 0.9748
- eval_runtime: 1372.3143
- eval_samples_per_second: 49.825
- eval_steps_per_second: 3.114
- epoch: 0.83
- step: 8000
Training procedure
Download dataset -> get text only and save to fr.tsv -> train BPE Tokenizer -> training
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 64
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
- mixed_precision_training: Native AMP
Framework versions
- Transformers 4.36.0
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.15.0
- Downloads last month
- 0