deberta-v3-large_20

This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1346
• F1 Micro: 0.3303
• F1 Macro: 0.0103
• Exact Match: 0.0

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: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Micro	F1 Macro	Exact Match
0.0906	1.0	656	0.1086	0.5700	0.0268	0.0
0.0737	2.0	1312	0.0961	0.5972	0.0333	0.0
0.0733	3.0	1968	0.0963	0.5962	0.0333	0.0
0.0779	4.0	2624	0.1057	0.5765	0.0274	0.0
0.0776	5.0	3280	0.1082	0.5561	0.0282	0.0
0.0765	6.0	3936	0.1101	0.5552	0.0272	0.0
0.0768	7.0	4592	0.1099	0.5753	0.0273	0.0
0.0748	8.0	5248	0.1095	0.5796	0.0278	0.0
0.0756	9.0	5904	0.1117	0.5551	0.0245	0.0
0.0764	10.0	6560	0.1181	0.4265	0.0158	0.0
0.0769	11.0	7216	0.1089	0.5551	0.0245	0.0
0.0771	12.0	7872	0.1281	0.3938	0.0134	0.0
0.0761	13.0	8528	0.1348	0.3230	0.0102	0.0
0.0760	14.0	9184	0.1193	0.5551	0.0245	0.0
0.0774	15.0	9840	0.1327	0.3303	0.0103	0.0
0.0764	16.0	10496	0.1360	0.2547	0.0072	0.0
0.0754	17.0	11152	0.1285	0.3798	0.0130	0.0
0.0764	18.0	11808	0.1348	0.3303	0.0103	0.0
0.0772	19.0	12464	0.1327	0.3303	0.0103	0.0
0.0758	20.0	13120	0.1346	0.3303	0.0103	0.0

Framework versions

• Transformers 5.2.0
• Pytorch 2.10.0+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2