Improved-MARBERT-twitter-sentiment-Twitter

This model is a fine-tuned version of UBC-NLP/MARBERTv2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7706
• Accuracy: 0.86

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5838	0.55	50	0.6058	0.71
0.3547	1.1	100	0.3887	0.83
0.2792	1.65	150	0.3479	0.85
0.1929	2.2	200	0.3596	0.87
0.1725	2.75	250	0.5874	0.8
0.1342	3.3	300	0.6560	0.81
0.1179	3.85	350	0.5146	0.85
0.079	4.4	400	0.6173	0.83
0.0928	4.95	450	0.7558	0.81
0.0425	5.49	500	1.0791	0.77
0.0609	6.04	550	0.7408	0.85
0.0328	6.59	600	0.8294	0.82
0.0531	7.14	650	0.6755	0.86
0.0342	7.69	700	0.6880	0.86
0.0263	8.24	750	0.7326	0.86
0.0147	8.79	800	0.8116	0.85
0.0169	9.34	850	0.8261	0.86
0.0118	9.89	900	0.7473	0.88
0.0087	10.44	950	0.7959	0.86
0.0051	10.99	1000	0.8585	0.85
0.0086	11.54	1050	0.8035	0.87
0.0076	12.09	1100	0.8838	0.84
0.0048	12.64	1150	0.8124	0.87
0.0095	13.19	1200	0.9262	0.85
0.0024	13.74	1250	0.8280	0.86
0.0109	14.29	1300	0.7895	0.87
0.0038	14.84	1350	0.7706	0.86

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.7
• Tokenizers 0.14.1