Fetis789/deberta_qqp

shad_qpp_week_4_model

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2422
• Model Preparation Time: 0.0033
• Accuracy: 0.9230
• Precision: 0.8817
• Recall: 0.9135
• F1: 0.8973

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: 32
• 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: 3400
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Accuracy	Precision	Recall	F1
0.3286	0.4397	5000	0.2468	0.0033	0.8953	0.8611	0.8533	0.8572
0.2439	0.8794	10000	0.2343	0.0033	0.9021	0.8307	0.9220	0.8739
0.1938	1.3191	15000	0.2153	0.0033	0.9134	0.8738	0.8937	0.8837
0.1749	1.7589	20000	0.2150	0.0033	0.9181	0.8828	0.8966	0.8896
0.1503	2.1986	25000	0.2329	0.0033	0.9202	0.8896	0.8943	0.8919
0.1225	2.6383	30000	0.2293	0.0033	0.9221	0.8789	0.9145	0.8963
0.1225	3.0	34113	0.2422	0.0033	0.9230	0.8817	0.9135	0.8973

Framework versions

• Transformers 5.16.1
• Pytorch 2.11.0+cu128
• Datasets 5.0.1
• Tokenizers 0.23.1