NLI-stance-immigration-burnham-finetuned-german

This model is a fine-tuned version of mlburnham/Political_DEBATE_large_v1.0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8002
• Accuracy: 0.7019
• F1 Macro: 0.6509
• Accuracy Balanced: 0.6591
• F1 Micro: 0.7019
• Precision Macro: 0.6544
• Recall Macro: 0.6591
• Precision Micro: 0.7019
• Recall Micro: 0.7019

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: 4
• eval_batch_size: 40
• seed: 1234
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.25
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	Accuracy Balanced	F1 Micro	Precision Macro	Recall Macro	Precision Micro	Recall Micro
No log	1.0	426	0.9146	0.7692	0.6634	0.6591	0.7692	0.7136	0.6591	0.7692	0.7692
0.9641	2.0	852	1.4334	0.6635	0.6198	0.6420	0.6635	0.6483	0.6420	0.6635	0.6635
0.296	3.0	1278	1.6905	0.6635	0.6215	0.6420	0.6635	0.6324	0.6420	0.6635	0.6635
0.1052	4.0	1704	1.7606	0.7115	0.6595	0.6667	0.7115	0.6632	0.6667	0.7115	0.7115
0.0141	5.0	2130	1.8002	0.7019	0.6509	0.6591	0.7019	0.6544	0.6591	0.7019	0.7019

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.19.1