Malaya Speech Mrbrown Finetune1

🚀 malaya-speech_Mrbrown_finetune1

This model is a fine - tuned version of [malay - huggingface/wav2vec2 - xls - r - 300m - mixed](https://huggingface.co/malay - huggingface/wav2vec2 - xls - r - 300m - mixed) on the uob_singlish dataset. It aims to achieve better performance in speech - related tasks on the specific dataset.

🚀 Quick Start

This section provides a brief overview of the model and its performance. The model was fine - tuned on a self - made dataset created by slicing the audio from "https://www.youtube.com/watch?v=a2ZOTD3R7JI" and writing corresponding transcripts, with a total duration of 4 minutes. However, the fine - tuning results were quite poor.

Evaluation Results

It achieves the following results on the evaluation set:

• Loss: 3.8458
• Wer: 1.01

🔧 Technical Details

Training Hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.01
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 100
• mixed_precision_training: Native AMP

Training Results

Framework Versions

• Transformers 4.11.3
• Pytorch 1.10.0+cu113
• Datasets 1.18.3
• Tokenizers 0.10.3

📚 Documentation

Model Description

More information needed

Intended Uses & Limitations

More information needed

Training and Evaluation Data

More information needed

Analysis of Poor Fine - Tuning Results

The poor fine - tuning results may imply that the training/fine - tuning dataset must be of high quality and at least several hours long. Another possible reason could be that the learning rate was set too high (0.01). Further research is needed to identify the important factors.