<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

ibo_xlmr

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

• eval_loss: 0.9762
• eval_runtime: 31.9667
• eval_samples_per_second: 32.471
• eval_steps_per_second: 4.067
• step: 0

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: 5
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 10
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Framework versions

• Transformers 4.21.1
• Pytorch 1.7.1+cu110
• Datasets 1.16.1
• Tokenizers 0.12.1