Edit model card

albert-base-ours-run-5

This model is a fine-tuned version of albert-base-v2 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 2.6151
  • Accuracy: 0.675
  • Precision: 0.6356
  • Recall: 0.6360
  • F1: 0.6356

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: 1e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1
0.9766 1.0 200 0.8865 0.645 0.5935 0.5872 0.5881
0.7725 2.0 400 1.0650 0.665 0.7143 0.5936 0.5556
0.6018 3.0 600 0.8558 0.7 0.6637 0.6444 0.6456
0.3838 4.0 800 0.9796 0.67 0.6220 0.6219 0.6218
0.2135 5.0 1000 1.4533 0.675 0.6611 0.5955 0.6055
0.1209 6.0 1200 1.4688 0.67 0.6392 0.6474 0.6398
0.072 7.0 1400 1.8395 0.695 0.6574 0.6540 0.6514
0.0211 8.0 1600 2.0849 0.7 0.6691 0.6607 0.6603
0.0102 9.0 1800 2.3042 0.695 0.6675 0.6482 0.6533
0.0132 10.0 2000 2.2390 0.685 0.6472 0.6423 0.6439
0.004 11.0 2200 2.3779 0.68 0.6435 0.6481 0.6443
0.0004 12.0 2400 2.4575 0.675 0.6397 0.6352 0.6357
0.0003 13.0 2600 2.4676 0.675 0.6356 0.6360 0.6356
0.0003 14.0 2800 2.5109 0.68 0.6427 0.6424 0.6422
0.0002 15.0 3000 2.5470 0.675 0.6356 0.6360 0.6356
0.0002 16.0 3200 2.5674 0.675 0.6356 0.6360 0.6356
0.0001 17.0 3400 2.5889 0.685 0.6471 0.6488 0.6474
0.0001 18.0 3600 2.6016 0.675 0.6356 0.6360 0.6356
0.0001 19.0 3800 2.6108 0.675 0.6356 0.6360 0.6356
0.0001 20.0 4000 2.6151 0.675 0.6356 0.6360 0.6356

Framework versions

  • Transformers 4.25.1
  • Pytorch 1.13.0+cu116
  • Tokenizers 0.13.2
Downloads last month
20
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for SkyR/albert-base-ours-run-5

Finetuned
(166)
this model