---
library_name: transformers
license: mit
base_model: FacebookAI/roberta-large
tags:
- generated_from_trainer
datasets:
- social_i_qa
metrics:
- accuracy
model-index:
- name: test-glue
  results: []
---

<!-- 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. -->

# test-glue

This model is a fine-tuned version of [FacebookAI/roberta-large](https://huggingface.co/FacebookAI/roberta-large) on the social_i_qa dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6767
- Accuracy: 0.7897

## 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: 1e-05
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.98) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 400
- num_epochs: 3
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Accuracy |
|:-------------:|:------:|:----:|:---------------:|:--------:|
| 0.4909        | 1.0    | 2089 | 0.5585          | 0.7769   |
| 0.3564        | 2.0    | 4178 | 0.5851          | 0.7876   |
| 0.2002        | 2.9990 | 6264 | 0.6767          | 0.7897   |


### Framework versions

- Transformers 4.52.0.dev0
- Pytorch 2.6.0+cu124
- Datasets 3.5.2.dev0
- Tokenizers 0.21.1