taskydata/deberta-v3-base_10xp3nirstbbflanse_5xc4

Hyperparameters:

• learning rate: 2e-5
• weight decay: 0.01
• per_device_train_batch_size: 8
• per_device_eval_batch_size: 8
• gradient_accumulation_steps:1
• eval steps: 24000
• max_length: 512
• num_epochs: 2
• hidden_dropout_prob: 0.3
• attention_probs_dropout_prob: 0.25

Dataset version:

• taskydata/10xp3nirstbbflanse_5xc4

Checkpoint:

• 48000 steps

Results on Validation set:

Step	Training Loss	Validation Loss	Accuracy	Precision	Recall	F1
24000	0.052000	0.071572	0.988261	0.999752	0.987852	0.993767
48000	0.015100	0.026952	0.995925	0.999564	0.996132	0.997846

Wandb logs:

• https://wandb.ai/manandey/huggingface/runs/2vh7iwi6?workspace=user-manandey