distilbert-sentiment-sft-cft

This model is a fine-tuned version of wsklee/distilbert-sentiment-sft on the None dataset. It achieves the following results on the evaluation set:

• Accuracy: 0.9960
• Pos Similarity: 0.9757
• Neg Similarity: 0.3175
• F1: 0.9980
• Precision: 1.0
• Recall: 0.9960
• Loss: 3.4705

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: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Accuracy	Pos Similarity	Neg Similarity	F1	Precision	Recall	Validation Loss
3.8387	0.5682	200	0.9928	0.9855	0.1089	0.9964	1.0	0.9928	3.5018
3.5835	1.1364	400	0.9896	0.9875	0.4204	0.9948	1.0	0.9896	3.4739
3.525	1.7045	600	0.9944	0.9903	0.3887	0.9972	1.0	0.9944	3.4476
3.5044	2.2727	800	0.9952	0.9878	0.4102	0.9976	1.0	0.9952	3.4468
3.4916	2.8409	1000	0.9944	0.9902	0.4468	0.9972	1.0	0.9944	3.4431
3.4697	3.4091	1200	0.9952	0.9814	0.3437	0.9976	1.0	0.9952	3.4546
3.4629	3.9773	1400	0.9960	0.9795	0.3308	0.9980	1.0	0.9960	3.4592
3.4465	4.5455	1600	0.9960	0.9757	0.3175	0.9980	1.0	0.9960	3.4705

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1
• Datasets 3.1.0
• Tokenizers 0.20.3