deberta-v3-base_smcalflow-classifier_20

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

• Loss: 0.0718
• F1 Micro: 0.8731
• F1 Macro: 0.2704
• Exact Match: 0.3

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
0.1028	1.0	656	0.1133	0.5700	0.0268	0.0
0.0761	2.0	1312	0.0969	0.5976	0.0334	0.0
0.0568	3.0	1968	0.0790	0.7348	0.0651	0.0042
0.0404	4.0	2624	0.0710	0.7911	0.1132	0.0653
0.0319	5.0	3280	0.0661	0.8167	0.1316	0.0972
0.0268	6.0	3936	0.0660	0.8312	0.1582	0.1069
0.0250	7.0	4592	0.0676	0.8434	0.1738	0.1278
0.0218	8.0	5248	0.0680	0.8481	0.1926	0.1431
0.0189	9.0	5904	0.0738	0.8458	0.1938	0.1653
0.0180	10.0	6560	0.0696	0.8538	0.2105	0.1819
0.0167	11.0	7216	0.0688	0.8626	0.2223	0.2208
0.0158	12.0	7872	0.0667	0.8680	0.2337	0.2333
0.0142	13.0	8528	0.0714	0.8657	0.2435	0.2542
0.0137	14.0	9184	0.0708	0.8695	0.2378	0.2639
0.0124	15.0	9840	0.0700	0.8697	0.2515	0.275
0.0120	16.0	10496	0.0719	0.8720	0.2608	0.2875
0.0116	17.0	11152	0.0716	0.8711	0.2598	0.2847
0.0109	18.0	11808	0.0724	0.8722	0.2610	0.2861
0.0112	19.0	12464	0.0715	0.8735	0.2680	0.3
0.0111	20.0	13120	0.0718	0.8731	0.2704	0.3

Framework versions

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