COPAS-whisper-lg-3-Nov3

sqrk/COPAS-whisper-lg-3-Nov3

COPAS-whisper-lg-3-Nov3 is a machine learning model from sqrk. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.

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

Downloads · 30 days

51% of all-time downloads

All-time downloads

Public

Parameters

1.5B

185 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors6.2 GB · 100%

At a glance

License: apache-2.0
Model type: whisper
Access: Public
Created: Nov 28, 2024
Updated: Nov 29, 2024
SHA: 49e1d125

Base models

openai/whisper-large-v3

Model card

From the Hugging Face model README

COPAS-whisper-lg-3-Nov3

This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0885
Wer: 13.9328

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: 1e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 100
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8006	1.0638	100	0.3410	23.5700
0.1853	2.1277	200	0.1805	19.6305
0.0717	3.1915	300	0.1121	17.7164
0.0313	4.2553	400	0.1272	20.3428
0.0206	5.3191	500	0.1029	17.0710
0.0112	6.3830	600	0.1037	17.9835
0.0058	7.4468	700	0.0975	16.7371
0.0046	8.5106	800	0.0938	15.4240
0.0042	9.5745	900	0.0882	14.4002
0.0053	10.6383	1000	0.0887	16.2252
0.0062	11.7021	1100	0.0938	16.2252
0.0048	12.7660	1200	0.0859	16.1807
0.004	13.8298	1300	0.0818	16.2698
0.0016	14.8936	1400	0.0770	14.1554
0.0004	15.9574	1500	0.0794	14.4892
0.0003	17.0213	1600	0.0785	15.1124
0.0001	18.0851	1700	0.0798	13.9550
0.0001	19.1489	1800	0.0805	13.5544
0.0001	20.2128	1900	0.0808	13.5989
0.0001	21.2766	2000	0.0810	13.6212
0.0001	22.3404	2100	0.0814	14.1776
0.0001	23.4043	2200	0.0817	14.1776
0.0001	24.4681	2300	0.0820	14.5337
0.0	25.5319	2400	0.0822	14.0663
0.0	26.5957	2500	0.0823	14.0441
0.0	27.6596	2600	0.0826	14.0441
0.0	28.7234	2700	0.0828	14.0441
0.0	29.7872	2800	0.0830	14.0441
0.0	30.8511	2900	0.0833	13.3986
0.0	31.9149	3000	0.0834	13.3986
0.0	32.9787	3100	0.0837	13.3986
0.0	34.0426	3200	0.0839	13.4209
0.0	35.1064	3300	0.0841	13.4209
0.0	36.1702	3400	0.0842	13.4209
0.0	37.2340	3500	0.0844	13.4209
0.0	38.2979	3600	0.0846	14.2221
0.0	39.3617	3700	0.0849	14.2221
0.0	40.4255	3800	0.0851	14.1999
0.0	41.4894	3900	0.0853	14.1554
0.0	42.5532	4000	0.0854	14.1554
0.0	43.6170	4100	0.0856	14.1554
0.0	44.6809	4200	0.0857	14.1554
0.0	45.7447	4300	0.0860	14.1331
0.0	46.8085	4400	0.0862	14.1554
0.0	47.8723	4500	0.0863	14.1554
0.0	48.9362	4600	0.0866	14.1331
0.0	50.0	4700	0.0867	14.1331
0.0	51.0638	4800	0.0868	14.1331
0.0	52.1277	4900	0.0869	14.1331
0.0	53.1915	5000	0.0870	14.1999
0.0	54.2553	5100	0.0873	14.1999
0.0	55.3191	5200	0.0875	14.2221
0.0	56.3830	5300	0.0876	14.2221
0.0	57.4468	5400	0.0879	13.9550
0.0	58.5106	5500	0.0880	13.9550
0.0	59.5745	5600	0.0881	13.9550
0.0	60.6383	5700	0.0883	13.9773
0.0	61.7021	5800	0.0885	13.9328

Framework versions

Transformers 4.43.4
Pytorch 2.4.1
Datasets 3.0.0
Tokenizers 0.19.1

Type: whisper
License: apache-2.0
Created: Nov 28, 2024
Updated: Nov 29, 2024