AlekseyKorshuk's picture
update model card README.md
6017b4a
metadata
license: other
tags:
  - generated_from_trainer
datasets:
  - AlekseyKorshuk/dalio-book-handwritten-io-sorted
metrics:
  - accuracy
model-index:
  - name: 6.7b-dalio-book-handwritten-io-constant-3e-7
    results:
      - task:
          name: Causal Language Modeling
          type: text-generation
        dataset:
          name: AlekseyKorshuk/dalio-book-handwritten-io-sorted
          type: AlekseyKorshuk/dalio-book-handwritten-io-sorted
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.30175150519978106

6.7b-dalio-book-handwritten-io-constant-3e-7

This model is a fine-tuned version of facebook/opt-6.7b on the AlekseyKorshuk/dalio-book-handwritten-io-sorted dataset. It achieves the following results on the evaluation set:

  • Loss: 2.4629
  • Accuracy: 0.3018

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: 3e-07
  • train_batch_size: 1
  • eval_batch_size: 1
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 8
  • total_train_batch_size: 8
  • total_eval_batch_size: 8
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: constant
  • num_epochs: 1.0

Training results

Training Loss Epoch Step Validation Loss Accuracy
2.6398 0.11 6 2.5059 0.2987
2.5823 0.21 12 2.5039 0.2988
2.6128 0.32 18 2.4980 0.2991
2.5775 0.43 24 2.4922 0.2995
2.527 0.54 30 2.4863 0.2999
2.5752 0.64 36 2.4805 0.3003
2.5131 0.75 42 2.4746 0.3008
2.4436 0.86 48 2.4688 0.3014
2.5114 0.96 54 2.4629 0.3018

Framework versions

  • Transformers 4.25.0.dev0
  • Pytorch 1.12.1+cu113
  • Datasets 2.3.2
  • Tokenizers 0.12.1