sentiment-analysis-roberta-base-V1.3

This model is a fine-tuned version of FacebookAI/roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9733
• Accuracy: 0.7033
• Precision Macro: 0.7074
• Recall Macro: 0.6760
• F1 Macro: 0.6838
• Precision Weighted: 0.7054
• Recall Weighted: 0.7033
• F1 Weighted: 0.6967

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 8
• seed: 3407
• optimizer: Use OptimizerNames.ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 50
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	Precision Weighted	Recall Weighted	F1 Weighted
0.0883	0.4545	20	1.3057	0.6752	0.6693	0.6745	0.6689	0.6878	0.6752	0.6785
0.0096	0.9091	40	1.6201	0.7033	0.6995	0.6998	0.6959	0.7108	0.7033	0.7035
0.2317	1.3636	60	1.7050	0.7340	0.7271	0.7238	0.7246	0.7338	0.7340	0.7331
0.0225	1.8182	80	1.6905	0.7110	0.7077	0.6911	0.6963	0.7097	0.7110	0.7074
0.2277	2.2727	100	1.8978	0.7059	0.7028	0.6830	0.6864	0.7056	0.7059	0.7001
0.3325	2.7273	120	1.9733	0.7033	0.7074	0.6760	0.6838	0.7054	0.7033	0.6967

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1