---
library_name: transformers
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
model-index:
- name: whisper-small-sc
  results: []
datasets:
- janaab/supreme-court-speech
language:
- en
metrics:
- wer
pipeline_tag: automatic-speech-recognition
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# whisper-small-sc

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on [janaab/supreme-court-speech](https://huggingface.co/janaab/supreme-court-speech) dataset.
It achieves the following results on the evaluation set:
- eval_loss: 0.3354
- eval_wer_ortho: 11.0780
- eval_wer: 10.5653
- eval_runtime: 1059.0881
- eval_samples_per_second: 4.216
- eval_steps_per_second: 0.264
- epoch: 6.9337
- step: 2250

## 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: 5e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine_with_restarts
- lr_scheduler_warmup_steps: 50
- training_steps: 4000
- mixed_precision_training: Native AMP

### Framework versions

- Transformers 4.44.2
- Pytorch 2.4.0
- Datasets 2.21.0
- Tokenizers 0.19.1