Visualize in Weights & Biases

mms-300m-turkish

This model is a fine-tuned version of mms-meta/mms-zeroshot-300m on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4082
  • Wer: 0.5510
  • Cer: 0.1163

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: 0.01
  • train_batch_size: 4
  • eval_batch_size: 8
  • seed: 45
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: warmup_stable_decay
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss Wer Cer
No log 0 0 21.7600 1.0 2.7817
0.6123 0.16 1000 0.4738 0.5799 0.1275
0.5963 0.32 2000 0.4366 0.5582 0.1195
0.5578 0.48 3000 0.4301 0.5731 0.1205
0.5414 0.64 4000 0.4181 0.5598 0.1175
0.5381 0.8 5000 0.4106 0.5535 0.1168
0.5366 0.96 6000 0.4082 0.5510 0.1163

Framework versions

  • Transformers 4.49.0
  • Pytorch 2.6.0+cu124
  • Datasets 3.3.2
  • Tokenizers 0.21.0
Downloads last month
464
Safetensors
Model size
317M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API: The model has no pipeline_tag.

Model tree for Kellner/mms-300m-turkish

Finetuned
(11)
this model