ikema-asr

ctaguchi/ikema-asr

ikema-asr is a automatic speech recognition model from ctaguchi. 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

3% of all-time downloads

All-time downloads

197

Public

Parameters

316M

56.8 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: Oct 8, 2025
Updated: Oct 8, 2025
SHA: 4b6463ff

Base models

facebook/wav2vec2-xls-r-300m

Model card

From the Hugging Face model README

ikema-asr

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 4.4069
Cer: 0.7119

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: 0.0003
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.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: 100
num_epochs: 50
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
11.3881	1.1117	100	3.8517	0.9927
3.9077	2.2235	200	3.6587	0.9925
3.8977	3.3352	300	3.6576	0.9927
3.8192	4.4469	400	3.6630	0.9929
3.7035	5.5587	500	3.6788	0.9789
3.6207	6.6704	600	3.6348	0.9544
3.5701	7.7821	700	3.5943	0.9705
3.5264	8.8939	800	3.5571	0.9585
3.3878	10.0	900	3.5424	0.9619
3.2684	11.1117	1000	3.3908	0.9472
3.0432	12.2235	1100	3.5293	0.8914
2.7166	13.3352	1200	2.8965	0.8707
2.372	14.4469	1300	2.9027	0.8354
2.1382	15.5587	1400	2.7674	0.8356
1.9442	16.6704	1500	2.9376	0.8006
1.7881	17.7821	1600	2.7831	0.7228
1.6766	18.8939	1700	2.8390	0.7168
1.5416	20.0	1800	3.0158	0.7305
1.4251	21.1117	1900	2.9839	0.6959
1.32	22.2235	2000	2.9795	0.7089
1.2355	23.3352	2100	3.1453	0.7217
1.1704	24.4469	2200	3.0608	0.7028
1.0789	25.5587	2300	3.0710	0.7133
1.0109	26.6704	2400	3.1124	0.7096
0.9362	27.7821	2500	3.4385	0.7221
0.8671	28.8939	2600	3.3170	0.7152
0.831	30.0	2700	3.2871	0.7012
0.7467	31.1117	2800	3.5183	0.7048
0.7128	32.2235	2900	3.3886	0.7088
0.6815	33.3352	3000	3.7318	0.7230
0.6452	34.4469	3100	3.7760	0.7320
0.6052	35.5587	3200	3.8134	0.7281
0.5779	36.6704	3300	3.7409	0.6809
0.5343	37.7821	3400	3.8252	0.6907
0.5056	38.8939	3500	3.9913	0.6964
0.5017	40.0	3600	4.1250	0.7000
0.4592	41.1117	3700	4.2096	0.7221
0.4478	42.2235	3800	4.2160	0.7031
0.4172	43.3352	3900	4.2452	0.7085
0.4018	44.4469	4000	4.3075	0.7155
0.3854	45.5587	4100	4.3594	0.7095
0.3771	46.6704	4200	4.3663	0.7133
0.3553	47.7821	4300	4.4198	0.7135
0.3565	48.8939	4400	4.4136	0.7108

Framework versions

Transformers 4.51.2
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1

Task: Automatic Speech Recognition
Library: transformers
Type: wav2vec2
License: apache-2.0
Created: Oct 8, 2025
Updated: Oct 8, 2025