visobert-human-tl-seed-1337

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

• Loss: 0.3650
• Accuracy: 0.8656
• Precision: 0.7112
• Recall: 0.5851
• F1: 0.6160

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• 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_steps: 500
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
No log	1.0	346	0.4107	0.8522	0.7081	0.5036	0.5214
0.4895	2.0	692	0.3926	0.8578	0.6850	0.5391	0.5648
0.3674	3.0	1038	0.3935	0.8559	0.6970	0.5343	0.5612
0.3674	4.0	1384	0.3842	0.8604	0.7129	0.5493	0.5819
0.3621	5.0	1730	0.3806	0.8608	0.7115	0.5538	0.5853
0.3533	6.0	2076	0.3853	0.8623	0.7161	0.5674	0.5921
0.3533	7.0	2422	0.3767	0.8608	0.7020	0.5558	0.5949
0.3511	8.0	2768	0.3780	0.8634	0.7105	0.5588	0.5920
0.3465	9.0	3114	0.3715	0.8604	0.6863	0.5575	0.5902
0.3465	10.0	3460	0.3801	0.8593	0.7125	0.5558	0.5771
0.3518	11.0	3806	0.3695	0.8626	0.7088	0.5673	0.6058
0.3415	12.0	4152	0.3734	0.8615	0.6970	0.5842	0.6075
0.3415	13.0	4498	0.3702	0.8679	0.7294	0.5687	0.6110
0.3427	14.0	4844	0.3684	0.8656	0.7118	0.5692	0.6078
0.335	15.0	5190	0.3642	0.8645	0.7072	0.5731	0.6084
0.3404	16.0	5536	0.3693	0.8660	0.7346	0.5577	0.5952
0.3404	17.0	5882	0.3650	0.8656	0.7112	0.5851	0.6160
0.3287	18.0	6228	0.3689	0.8630	0.7187	0.5699	0.5946
0.3299	19.0	6574	0.3636	0.8660	0.7064	0.5817	0.6116
0.3299	20.0	6920	0.3620	0.8668	0.7256	0.5792	0.6099
0.3215	21.0	7266	0.3635	0.8671	0.7338	0.5758	0.6045
0.3249	22.0	7612	0.3619	0.8668	0.7327	0.5732	0.6021

Framework versions

• Transformers 4.51.1
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0