sarcasm-RuBERT

<!-- 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_trainer_16_05

This model is a fine-tuned version of sismetanin/rubert-ru-sentiment-rusentiment on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2726
• Accuracy: 0.8545
• Precision: 0.8305
• Recall: 0.8909
• F1: 0.8596

Model description

A fine-tuned version of Ru-BERT, finetuned on sentiment analysis tasks. Two class sarcasm detection, trained on a manually annotated dataset of russian negative product reviews (3 annotators), partially tested on an ambiguious data.

Intended uses & limitations

More information needed

Training and evaluation data

A manually annotated dataset of russian negative product reviews (3 annotators)

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 29
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.1