Edit model card

Vreme Model Tiny - finetuned on horoscope

This model is a fine-tuned version of iulik-udrik/vreme_model_tiny on the Horoscop Neti Sandu dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4493
  • Wer: 23.3390

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: 16
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • training_steps: 4000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.0345 9.71 1000 0.3666 23.7527
0.0022 19.42 2000 0.4232 22.7306
0.0011 29.13 3000 0.4421 23.1930
0.0009 38.83 4000 0.4493 23.3390

Framework versions

  • Transformers 4.39.2
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2
Downloads last month
5
Safetensors
Model size
37.8M params
Tensor type
F32
·
Invalid base_model specified in model card metadata. Needs to be a model id from hf.co/models.

Evaluation results