metadata

tags:
  - generated_from_trainer
datasets:
  - Graphcore/wikipedia-bert-512
model-index:
  - name: output-pretrain-groupbert-base-phase2
    results: []

output-pretrain-groupbert-base-phase2

This model was trained from scratch on the Graphcore/wikipedia-bert-512 dataset.

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: 0.01
train_batch_size: 2
eval_batch_size: 1
seed: 42
distributed_type: IPU
gradient_accumulation_steps: 2048
total_train_batch_size: 16384
total_eval_batch_size: 20
optimizer: LAMB
lr_scheduler_type: linear
training_steps: 1
training precision: Mixed Precision

Training results

Framework versions

Transformers 4.20.1
Pytorch 1.10.0+cpu
Datasets 2.2.2
Tokenizers 0.12.1