wav2vec2-common_voice-hi-mms-demo

hanasim/wav2vec2-common_voice-hi-mms-demo

wav2vec2-common_voice-hi-mms-demo is a automatic speech recognition model from hanasim. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as cc-by-nc-4.0.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

4% of all-time downloads

All-time downloads

275

Public

Parameters

965M

69.5 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors3.9 GB · 100%

At a glance

Task: Automatic Speech Recognition
Library: transformers
License: cc-by-nc-4.0
Model type: wav2vec2
Access: Public
Created: Jan 27, 2024
Updated: Jan 27, 2024
SHA: 7bc4bab4

Base models

facebook/mms-1b-all

Datasets

common_voice_16_0

Model card

From the Hugging Face model README

wav2vec2-common_voice-hi-mms-demo

This model is a fine-tuned version of facebook/mms-1b-all on the MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - HI dataset. It achieves the following results on the evaluation set:

Loss: 0.2672
Wer: 0.2516

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: 0.001
train_batch_size: 8
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: 100
num_epochs: 4.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.11	100	0.4487	0.3565
No log	0.23	200	0.3544	0.3317
No log	0.34	300	0.3693	0.3088
No log	0.45	400	0.3404	0.3040
1.5084	0.56	500	0.3346	0.2995
1.5084	0.68	600	0.3411	0.2936
1.5084	0.79	700	0.3175	0.2887
1.5084	0.9	800	0.3159	0.2898
1.5084	1.02	900	0.3139	0.3045
0.3485	1.13	1000	0.3067	0.2958
0.3485	1.24	1100	0.2969	0.2767
0.3485	1.35	1200	0.2916	0.2714
0.3485	1.47	1300	0.2893	0.2663
0.3485	1.58	1400	0.3183	0.2985
0.3152	1.69	1500	0.2961	0.2688
0.3152	1.81	1600	0.2848	0.2665
0.3152	1.92	1700	0.2844	0.2656
0.3152	2.03	1800	0.2855	0.2707
0.3152	2.14	1900	0.2887	0.2686
0.3058	2.26	2000	0.2858	0.2657
0.3058	2.37	2100	0.2814	0.2629
0.3058	2.48	2200	0.2809	0.2633
0.3058	2.6	2300	0.2779	0.2613
0.3058	2.71	2400	0.2745	0.2581
0.2861	2.82	2500	0.2769	0.2618
0.2861	2.93	2600	0.2742	0.2576
0.2861	3.05	2700	0.2730	0.2575
0.2861	3.16	2800	0.2727	0.2564
0.2861	3.27	2900	0.2726	0.2563
0.2839	3.39	3000	0.2713	0.2576
0.2839	3.5	3100	0.2690	0.2537
0.2839	3.61	3200	0.2706	0.2540
0.2839	3.72	3300	0.2687	0.2542
0.2839	3.84	3400	0.2671	0.2521
0.2706	3.95	3500	0.2673	0.2522

Framework versions

Transformers 4.38.0.dev0
Pytorch 2.1.2+cu121
Datasets 2.16.1
Tokenizers 0.15.1

Task: Automatic Speech Recognition
Library: transformers
Type: wav2vec2
License: cc-by-nc-4.0
Languages: hi
Created: Jan 27, 2024
Updated: Jan 27, 2024