ModernBERT-base_nli

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

• Loss: 3.4416
• Accuracy: 0.5623
• Precision Macro: 0.5618
• Recall Macro: 0.5627
• F1 Macro: 0.5621
• F1 Weighted: 0.5617

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: 5e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	F1 Weighted
2.164	1.0	72	1.0434	0.4483	0.4472	0.4484	0.4398	0.4395
2.0623	2.0	144	0.9968	0.4984	0.5026	0.4994	0.4983	0.4978
1.8507	3.0	216	1.0155	0.5016	0.5522	0.5034	0.4808	0.4802
1.7076	4.0	288	0.9344	0.5721	0.5902	0.5738	0.5572	0.5563
1.4431	5.0	360	0.9258	0.5756	0.5770	0.5768	0.5719	0.5714
1.1592	6.0	432	1.0425	0.5738	0.5831	0.5740	0.5693	0.5691
0.6916	7.0	504	1.2622	0.5659	0.5711	0.5670	0.5640	0.5636
0.3547	8.0	576	1.7560	0.5455	0.5495	0.5452	0.5460	0.5459
0.2534	9.0	648	2.1882	0.5494	0.5620	0.5515	0.5409	0.5401
0.1018	10.0	720	2.3462	0.5645	0.5641	0.5652	0.5633	0.5630
0.0931	11.0	792	2.6256	0.5565	0.5619	0.5582	0.5483	0.5475
0.0504	12.0	864	2.7252	0.5552	0.5570	0.5557	0.5555	0.5551
0.0379	13.0	936	2.9577	0.5517	0.5518	0.5521	0.5518	0.5515
0.0111	14.0	1008	3.2048	0.5614	0.5621	0.5621	0.5609	0.5604
0.0018	15.0	1080	3.3005	0.5610	0.5621	0.5612	0.5616	0.5613
0.0003	16.0	1152	3.3958	0.5610	0.5602	0.5615	0.5605	0.5601
0.0001	17.0	1224	3.4259	0.5623	0.5617	0.5628	0.5620	0.5617
0.0001	18.0	1296	3.4368	0.5619	0.5613	0.5623	0.5616	0.5612
0.0001	19.0	1368	3.4412	0.5619	0.5614	0.5623	0.5616	0.5613
0.0001	20.0	1440	3.4416	0.5623	0.5618	0.5627	0.5621	0.5617

Framework versions

• Transformers 4.55.0
• Pytorch 2.7.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4