USS-reward-model-WRS

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

• Loss: 0.1904
• Mse: 0.2744
• Mae: 0.4282
• R2: -0.4226
• Spearman Correlation: 0.2760

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: 2
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 10
• total_train_batch_size: 20
• 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
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Mse	Mae	R2	Spearman Correlation
2008.1245	1.0	97	0.2061	0.4692	0.5698	-1.4325	0.1502
7.1771	2.0	194	0.3361	0.4943	0.5834	-1.5623	0.1917
7.7410	3.0	291	0.2937	0.3963	0.5103	-1.0543	0.2489
5.2149	4.0	388	0.2743	0.4378	0.5365	-1.2698	0.2246
3.6741	5.0	485	0.2663	0.3378	0.4603	-0.7511	0.2570
2.2179	6.0	582	0.2146	0.4209	0.5321	-1.1819	0.2639
1.2121	7.0	679	0.1904	0.2744	0.4282	-0.4226	0.2760
0.7297	8.0	776	0.1718	0.3017	0.4467	-0.5643	0.3221
0.4153	9.0	873	0.1359	0.3359	0.4754	-0.7415	0.3131
0.1252	10.0	970	0.1326	0.3632	0.4948	-0.8832	0.3159

Framework versions

• Transformers 5.9.0
• Pytorch 2.12.0+cu130
• Datasets 4.8.5
• Tokenizers 0.22.2