BERT-Base-SE2025T11A-eng-v0.5

This model is a fine-tuned version of bhadresh-savani/bert-base-uncased-emotion on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3877
• F1 Micro: 0.7444
• F1 Macro: 0.7028
• F1 Label Anger: 0.5570
• F1 Label Fear: 0.8087
• F1 Label Joy: 0.6915
• F1 Label Sad: 0.7317
• F1 Label Surprise: 0.725

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: 2
• eval_batch_size: 2
• 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: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Micro	F1 Macro	F1 Label Anger	F1 Label Fear	F1 Label Joy	F1 Label Sad	F1 Label Surprise
0.5761	0.1029	100	0.5207	0.5379	0.3482	0.0	0.7668	0.4559	0.0448	0.4737
0.5062	0.2058	200	0.4711	0.6262	0.4402	0.0	0.7759	0.2783	0.6496	0.4974
0.4609	0.3086	300	0.4505	0.6784	0.5705	0.2295	0.8	0.5860	0.6406	0.5962
0.4421	0.4115	400	0.4639	0.6516	0.5346	0.1818	0.7792	0.5325	0.6352	0.5446
0.4494	0.5144	500	0.4186	0.6535	0.5614	0.3235	0.8016	0.5967	0.5877	0.4974
0.4368	0.6173	600	0.4234	0.6832	0.5650	0.1455	0.8110	0.6102	0.6753	0.5829
0.4017	0.7202	700	0.4025	0.7185	0.6161	0.2105	0.8198	0.6602	0.6522	0.7376
0.4394	0.8230	800	0.4114	0.7039	0.6192	0.4474	0.8049	0.4580	0.656	0.7295
0.4148	0.9259	900	0.3776	0.7247	0.6752	0.5176	0.8109	0.6595	0.6723	0.7155
0.3777	1.0288	1000	0.3775	0.7366	0.6860	0.5053	0.8136	0.6627	0.7170	0.7313
0.2584	1.1317	1100	0.3924	0.7151	0.6459	0.4225	0.8106	0.6592	0.6280	0.7090
0.2816	1.2346	1200	0.3764	0.7292	0.6561	0.3714	0.8166	0.6595	0.7016	0.7313
0.2614	1.3374	1300	0.3825	0.7197	0.6533	0.3947	0.8106	0.6519	0.6667	0.7424
0.2536	1.4403	1400	0.3899	0.7327	0.6702	0.3947	0.8057	0.6974	0.7097	0.7436
0.2871	1.5432	1500	0.4037	0.7123	0.6396	0.4474	0.8182	0.5714	0.6824	0.6787
0.2981	1.6461	1600	0.3986	0.7028	0.6392	0.4368	0.8211	0.6742	0.5556	0.7085
0.2842	1.7490	1700	0.3978	0.7335	0.6627	0.4	0.828	0.6702	0.6840	0.7315
0.2599	1.8519	1800	0.3968	0.7353	0.6761	0.4578	0.8357	0.6839	0.6803	0.7229
0.2847	1.9547	1900	0.3981	0.7401	0.6706	0.4	0.8313	0.6845	0.6833	0.7538
0.2298	2.0576	2000	0.4003	0.7214	0.6678	0.4773	0.8220	0.6736	0.7025	0.6635
0.1892	2.1605	2100	0.4056	0.7396	0.6766	0.4471	0.8406	0.6667	0.7143	0.7143
0.1665	2.2634	2200	0.4141	0.7352	0.6628	0.3836	0.8323	0.6809	0.6752	0.7422
0.1565	2.3663	2300	0.4025	0.7342	0.6652	0.4211	0.8406	0.6514	0.6894	0.7236
0.162	2.4691	2400	0.4134	0.72	0.6462	0.3889	0.8290	0.6629	0.6867	0.6636
0.1689	2.5720	2500	0.4128	0.7366	0.6657	0.4110	0.8330	0.6667	0.7029	0.7149
0.1588	2.6749	2600	0.4174	0.7373	0.6699	0.4267	0.8295	0.6667	0.7029	0.7236
0.164	2.7778	2700	0.4122	0.7385	0.6721	0.4267	0.8343	0.6848	0.7004	0.7143
0.2007	2.8807	2800	0.4115	0.7397	0.6728	0.4156	0.8357	0.6845	0.7004	0.7280
0.1701	2.9835	2900	0.4133	0.7365	0.6695	0.4156	0.8347	0.6811	0.7	0.7160

Framework versions

• Transformers 4.46.0
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.20.1