README.md

metadata

license: mit
base_model: roberta-base
tags:
  - generated_from_trainer
model-index:
  - name: Models-RoBERTa-1704501009.345538
    results: []

Models-RoBERTa-1704501009.345538

This model is a fine-tuned version of roberta-base on an MNLI dataset. It achieves the following results on the evaluation set (1000 instances of MNLI validation matched):

• eval_loss: 0.2357
• eval_accuracy: 0.92
• eval_runtime: 7.4465
• eval_samples_per_second: 134.292
• eval_steps_per_second: 4.297
• epoch: 1.03
• step: 12597

Model description

The baseline NLI model is a fine-tuned version of roberta-base for Text Classifacation on the MNLI dataset , with entailments as label 0 and all others (neutral or contradiction) as label 1.

Two classes:

• entailment: 0

• non-entailment: 1

Intended uses & limitations

More information needed

Training and evaluation data

Model's performance on the validation sets:

MNLI: 92.07%
MNLI-mm: 92.09%

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0