USS-reward-model-utterance-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: 0.5460
• Mse: 0.2326
• Mae: 0.3471
• R2: -1.1275
• Spearman Correlation: -0.0277

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
46.7627	1.0	591	0.5256	0.4280	0.5110	-2.9143	-0.0244
19.9530	2.0	1182	0.5341	0.4089	0.5007	-2.7388	0.0013
5.2161	3.0	1773	0.5949	0.3394	0.4523	-2.1036	-0.0037
2.1788	4.0	2364	0.7364	0.3554	0.4675	-2.2503	-0.0119
1.3887	5.0	2955	0.5760	0.3011	0.4235	-1.7535	-0.0242
1.0354	6.0	3546	0.6582	0.2985	0.4157	-1.7300	-0.0264
0.5253	7.0	4137	0.4425	0.2575	0.3813	-1.3543	-0.0119
0.3577	8.0	4728	0.6235	0.2456	0.3625	-1.2457	-0.0354
0.2553	9.0	5319	0.5635	0.2417	0.3541	-1.2103	-0.0295
0.1649	10.0	5910	0.5460	0.2326	0.3471	-1.1275	-0.0277

Framework versions

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