my_awesome_mind_model Open-source Audio Classification Model - Achieving Nearly 60% Accuracy through Fine-tuning

My Awesome Mind Model

Developed by Krithika-p

An audio classification model fine-tuned based on facebook/wav2vec2-base, achieving 58.92% accuracy on the evaluation set

Audio Classification

Transformers

Open Source License:Apache-2.0 #Audio Classification #wav2vec2 Fine-tuning #Moderate Accuracy

Downloads 15

Release Time : 12/9/2023

Model Overview

This is a fine-tuned model for audio classification tasks, based on the wav2vec2 architecture, suitable for scenarios such as sound event detection

Model Features

Based on wav2vec2 Architecture

Utilizes the Facebook open-source wav2vec2-base model as the foundational architecture, providing excellent audio feature extraction capabilities

Fine-tuning Optimization

Enhanced classification performance on the target dataset through 50 epochs of meticulous fine-tuning

Model Capabilities

Audio Classification

Sound Event Detection

Use Cases

Sound Recognition

Environmental Sound Classification

Identify and classify various types of sounds in the environment

Achieved 58.92% accuracy on the test set

🚀 my_awesome_mind_model

This model is a fine - tuned version of facebook/wav2vec2-base on the audiofolder dataset. It offers a solution for audio classification tasks, leveraging the pre - trained capabilities of the base model and adapting them to the specific characteristics of the audiofolder dataset.

🚀 Quick Start

This model is a fine - tuned version of facebook/wav2vec2-base on the audiofolder dataset. It achieves the following results on the evaluation set:

Loss: 1.3338
Accuracy: 0.5892

📚 Documentation

Model Information

Property	Details
Model Type	Fine - tuned version of facebook/wav2vec2 - base
Training Data	audiofolder dataset
Metrics	Accuracy

Training Procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 3e - 05
train_batch_size: 32
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 128
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.7071	0.95	14	2.7063	0.0602
2.7033	1.97	29	2.7006	0.0645
2.6835	2.98	44	2.6793	0.0817
2.6551	4.0	59	2.5549	0.1699
2.5023	4.95	73	2.3970	0.2258
2.4257	5.97	88	2.3068	0.2495
2.2542	6.98	103	2.2121	0.2688
2.2419	8.0	118	2.1736	0.2731
2.1278	8.95	132	2.1675	0.2430
2.0592	9.97	147	2.1207	0.2796
1.9576	10.98	162	2.0662	0.2731
1.9023	12.0	177	1.9738	0.3312
1.8367	12.95	191	2.0420	0.2903
1.7822	13.97	206	2.0161	0.2860
1.6934	14.98	221	2.0215	0.2989
1.7093	16.0	236	1.9287	0.3290
1.6158	16.95	250	1.8138	0.3849
1.5879	17.97	265	1.8043	0.3871
1.5249	18.98	280	1.9117	0.3548
1.4821	20.0	295	1.7242	0.4215
1.4629	20.95	309	1.6981	0.4538
1.3847	21.97	324	1.6701	0.4516
1.3595	22.98	339	1.6891	0.4495
1.298	24.0	354	1.6321	0.4667
1.2479	24.95	368	1.5519	0.4989
1.2135	25.97	383	1.5477	0.4839
1.1833	26.98	398	1.5437	0.5032
1.1298	28.0	413	1.5425	0.5097
1.079	28.95	427	1.5076	0.5247
1.0709	29.97	442	1.5288	0.5140
1.0286	30.98	457	1.4497	0.5419
0.9896	32.0	472	1.4663	0.5355
0.9707	32.95	486	1.4683	0.5333
0.9443	33.97	501	1.4977	0.5226
0.8998	34.98	516	1.4178	0.5505
0.9048	36.0	531	1.4131	0.5462
0.8587	36.95	545	1.3791	0.5634
0.84	37.97	560	1.4036	0.5527
0.8155	38.98	575	1.4139	0.5505
0.8086	40.0	590	1.3993	0.5462
0.808	40.95	604	1.3325	0.5914
0.7929	41.97	619	1.3500	0.5806
0.7635	42.98	634	1.3471	0.5720
0.761	44.0	649	1.3636	0.5634
0.7456	44.95	663	1.3551	0.5828
0.75	45.97	678	1.3431	0.5849
0.7232	46.98	693	1.3338	0.5871
0.7625	47.46	700	1.3338	0.5892

Framework versions

Transformers 4.35.2
Pytorch 2.1.0+cu118
Datasets 2.15.0
Tokenizers 0.15.0

📄 License

This project is licensed under the Apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご