---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- timit_asr
model-index:
- name: remake_base
  results: []
---

<!-- 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. -->

# remake_base

This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the timit_asr dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3289
- Cer: 0.1196

## 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: 0.0001
- 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: 1000
- training_steps: 2500
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Cer    |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 5.891         | 0.8   | 200  | 3.4674          | 0.9746 |
| 3.0345        | 1.61  | 400  | 2.3321          | 0.6943 |
| 1.3335        | 2.41  | 600  | 0.6416          | 0.1832 |
| 0.6266        | 3.21  | 800  | 0.4364          | 0.1458 |
| 0.5244        | 4.02  | 1000 | 0.3783          | 0.1425 |
| 0.4244        | 4.82  | 1200 | 0.3599          | 0.1331 |
| 0.3897        | 5.62  | 1400 | 0.3361          | 0.1323 |
| 0.3254        | 6.43  | 1600 | 0.3336          | 0.1258 |
| 0.3007        | 7.23  | 1800 | 0.3346          | 0.1264 |
| 0.2719        | 8.03  | 2000 | 0.3167          | 0.1192 |
| 0.2417        | 8.84  | 2200 | 0.3272          | 0.1205 |
| 0.2253        | 9.64  | 2400 | 0.3289          | 0.1196 |


### Framework versions

- Transformers 4.17.0
- Pytorch 2.4.0
- Datasets 1.18.3
- Tokenizers 0.21.0