twitter-bitcoin-emotion-classification

This model is a fine-tuned version of vinai/bertweet-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7032
• Accuracy: 0.6561
• F1: 0.6572
• Macro F1: 0.6226
• Precision: 0.6630
• Recall: 0.6561

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Macro F1	Precision	Recall
2.9772	0.2092	500	2.6140	0.4598	0.3456	0.2565	0.2855	0.4598
2.0831	0.4185	1000	1.9186	0.5043	0.4434	0.4006	0.5917	0.5043
1.7199	0.6277	1500	1.6219	0.5694	0.5441	0.5039	0.5912	0.5694
1.56	0.8370	2000	1.4268	0.6031	0.5922	0.5572	0.6167	0.6031
1.3383	1.0460	2500	1.3762	0.6220	0.6180	0.5814	0.6289	0.6220
1.3473	1.2553	3000	1.3732	0.6188	0.6132	0.5774	0.6367	0.6188
1.1643	1.4645	3500	1.4119	0.6239	0.6247	0.5894	0.6391	0.6239
1.2617	1.6738	4000	1.3296	0.6195	0.6219	0.5792	0.6453	0.6195
1.2549	1.8830	4500	1.3182	0.6330	0.6351	0.5975	0.6460	0.6330
0.8836	2.0921	5000	1.4422	0.6247	0.6298	0.6006	0.6528	0.6247
0.8652	2.3013	5500	1.4682	0.6475	0.6488	0.6120	0.6563	0.6475
0.9502	2.5106	6000	1.5682	0.6402	0.6438	0.6128	0.6559	0.6402
0.8769	2.7198	6500	1.5807	0.6392	0.6435	0.6100	0.6583	0.6392
0.7853	2.9291	7000	1.5285	0.6389	0.6403	0.6077	0.6555	0.6389
0.6828	3.1381	7500	1.6189	0.6467	0.6499	0.6170	0.6588	0.6467
0.6566	3.3474	8000	1.4571	0.6530	0.6519	0.6140	0.6558	0.6530
0.6767	3.5566	8500	1.6671	0.6567	0.6585	0.6216	0.6613	0.6567
0.6806	3.7659	9000	1.5203	0.6518	0.6542	0.6195	0.6628	0.6518
0.7289	3.9751	9500	1.5431	0.6524	0.6568	0.6224	0.6669	0.6524
0.5355	4.1841	10000	1.6772	0.6501	0.6528	0.6185	0.6619	0.6501
0.5352	4.3934	10500	1.7032	0.6561	0.6572	0.6226	0.6630	0.6561
0.615	4.6026	11000	1.6694	0.6504	0.6527	0.6175	0.6608	0.6504
0.5843	4.8119	11500	1.6443	0.6541	0.6567	0.6213	0.6634	0.6541

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1