zeroshot-classification-de

This model is a fine-tuned version of MoritzLaurer/deberta-v3-base-zeroshot-v1.1-all-33 on the michaelp11/wiki-tags dataset. It achieves the following results on the evaluation set:

• Loss: 0.4552
• F1 Micro: 0.859
• Accuracy: 0.859
• Precision Micro: 0.859
• Recall Micro: 0.859

Intended uses & limitations

Zeroshot classification for german language - yay.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 32
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Macro	F1 Micro	Accuracy Balanced	Accuracy	Precision Macro	Recall Macro	Precision Micro	Recall Micro
No log	1.0	313	0.3407	0.8507	0.851	0.8509	0.851	0.8534	0.8509	0.851	0.851
0.373	2.0	626	0.3830	0.8600	0.86	0.8600	0.86	0.86	0.8600	0.86	0.86
0.373	3.0	939	0.4552	0.8589	0.859	0.8590	0.859	0.8596	0.8590	0.859	0.859

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1