twitter-roberta-base-sentiment-latest_08112024T124327

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

• Loss: 0.5159
• F1: 0.8599
• Learning Rate: 0.0

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 600
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Rate
No log	0.9942	86	1.7858	0.1609	0.0000
No log	2.0	173	1.6525	0.3252	0.0000
No log	2.9942	259	1.3699	0.4597	0.0000
No log	4.0	346	1.1227	0.5431	0.0000
No log	4.9942	432	1.0111	0.5780	0.0000
1.3954	6.0	519	0.9166	0.6216	0.0000
1.3954	6.9942	605	0.8749	0.6515	1e-05
1.3954	8.0	692	0.7692	0.7182	0.0000
1.3954	8.9942	778	0.7127	0.7426	0.0000
1.3954	10.0	865	0.6492	0.7694	0.0000
1.3954	10.9942	951	0.6837	0.7554	0.0000
0.6615	12.0	1038	0.5879	0.8077	0.0000
0.6615	12.9942	1124	0.5643	0.8239	0.0000
0.6615	14.0	1211	0.5588	0.8279	0.0000
0.6615	14.9942	1297	0.5126	0.8485	0.0000
0.6615	16.0	1384	0.5192	0.8532	0.0000
0.6615	16.9942	1470	0.5163	0.8555	0.0000
0.29	18.0	1557	0.5206	0.8571	6e-07
0.29	18.9942	1643	0.5152	0.8603	1e-07
0.29	19.8844	1720	0.5159	0.8599	0.0

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.19.1