xlsr-nm-nomimose

susmitabhatt/xlsr-nm-nomimose

xlsr-nm-nomimose is a automatic speech recognition model from susmitabhatt. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.

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

Downloads · 30 days

18% of all-time downloads

All-time downloads

Public

Parameters

315M

36.6 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors1.3 GB · 100%

At a glance

Task: Automatic Speech Recognition
Library: transformers
License: apache-2.0
Model type: wav2vec2
Access: Public
Created: Nov 13, 2024
Updated: Nov 13, 2024
SHA: ccd5ab15

Base models

facebook/wav2vec2-large-xlsr-53

Model card

From the Hugging Face model README

xlsr-nm-nomimose

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.9684
Wer: 0.4369

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: 0.0004
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 132
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.9556	3.3932	200	3.0924	1.0
3.024	6.7863	400	2.8991	0.9943
2.7554	10.1709	600	2.4531	1.0
2.1877	13.5641	800	1.6865	0.9181
1.247	16.9573	1000	1.1531	0.7247
0.7161	20.3419	1200	1.0237	0.6052
0.4613	23.7350	1400	0.9152	0.5631
0.3173	27.1197	1600	0.8917	0.5165
0.2399	30.5128	1800	0.8512	0.5256
0.1871	33.9060	2000	0.9078	0.4937
0.1536	37.2906	2200	0.9574	0.4972
0.1259	40.6838	2400	0.9938	0.4903
0.1095	44.0684	2600	1.0196	0.4994
0.0947	47.4615	2800	0.9235	0.4778
0.073	50.8547	3000	1.1352	0.4972
0.068	54.2393	3200	0.9595	0.4778
0.0562	57.6325	3400	1.0105	0.4710
0.0564	61.0171	3600	1.0297	0.4744
0.052	64.4103	3800	1.0371	0.4562
0.0371	67.8034	4000	1.0999	0.4733
0.034	71.1880	4200	1.0486	0.4699
0.039	74.5812	4400	0.9800	0.4585
0.031	77.9744	4600	0.9614	0.4494
0.0323	81.3590	4800	0.9838	0.4551
0.0229	84.7521	5000	1.0129	0.4334
0.0232	88.1368	5200	0.9266	0.4243
0.0144	91.5299	5400	0.9751	0.4334
0.0178	94.9231	5600	0.9619	0.4369
0.0147	98.3077	5800	0.9684	0.4369

Framework versions

Transformers 4.47.0.dev0
Pytorch 2.4.0
Datasets 3.0.1
Tokenizers 0.20.0

Task: Automatic Speech Recognition
Library: transformers
Type: wav2vec2
License: apache-2.0
Created: Nov 13, 2024
Updated: Nov 13, 2024