w2v-bert-odia_v1

cdactvm/w2v-bert-odia_v1

w2v-bert-odia_v1 is a automatic speech recognition model from cdactvm. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as mit.

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

Downloads · 30 days

13% of all-time downloads

All-time downloads

Public

Parameters

606M

160 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors2.4 GB · 100%

At a glance

Task: Automatic Speech Recognition
Library: transformers
License: mit
Model type: wav2vec2-bert
Access: Public
Created: Jul 24, 2024
Updated: Aug 5, 2024
SHA: 3dbfd561

Base models

facebook/w2v-bert-2.0

Spaces

cdactvm/demoASR

Model card

From the Hugging Face model README

w2v-bert-odia_v1

This model is a fine-tuned version of facebook/w2v-bert-2.0 on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.2688
Wer: 0.1951

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: 5e-05
train_batch_size: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 4
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.9674	0.0342	300	1.3305	0.7001
1.2476	0.0683	600	1.1660	0.5879
1.0692	0.1025	900	0.9110	0.4886
0.9443	0.1366	1200	0.7601	0.4727
0.8235	0.1708	1500	0.7761	0.3973
0.8155	0.2050	1800	0.7084	0.4022
0.767	0.2391	2100	0.6251	0.3756
0.7517	0.2733	2400	0.6125	0.3654
0.687	0.3075	2700	0.5848	0.3439
0.6509	0.3416	3000	0.5643	0.3282
0.6632	0.3758	3300	0.5509	0.3199
0.6108	0.4099	3600	0.5393	0.3341
0.5898	0.4441	3900	0.5223	0.3277
0.595	0.4783	4200	0.5199	0.3200
0.5644	0.5124	4500	0.5508	0.2919
0.5787	0.5466	4800	0.4994	0.3060
0.5752	0.5807	5100	0.4966	0.2997
0.5353	0.6149	5400	0.4731	0.3237
0.5473	0.6491	5700	0.4665	0.3062
0.5498	0.6832	6000	0.4890	0.2876
0.5146	0.7174	6300	0.4747	0.2926
0.5398	0.7516	6600	0.4581	0.2907
0.5154	0.7857	6900	0.4557	0.2995
0.5386	0.8199	7200	0.4515	0.2948
0.5037	0.8540	7500	0.4456	0.2961
0.5344	0.8882	7800	0.4509	0.2988
0.501	0.9224	8100	0.4436	0.2711
0.487	0.9565	8400	0.4233	0.2749
0.4692	0.9907	8700	0.4661	0.2532
0.462	1.0249	9000	0.4197	0.2723
0.4508	1.0590	9300	0.4316	0.2584
0.4702	1.0932	9600	0.4148	0.2689
0.4517	1.1273	9900	0.3950	0.2549
0.4408	1.1615	10200	0.4308	0.2551
0.4636	1.1957	10500	0.4033	0.2700
0.4583	1.2298	10800	0.4096	0.2556
0.4315	1.2640	11100	0.3883	0.2681
0.4172	1.2981	11400	0.3737	0.2529
0.4177	1.3323	11700	0.3992	0.2472
0.3975	1.3665	12000	0.3716	0.2485
0.4044	1.4006	12300	0.3853	0.2523
0.4497	1.4348	12600	0.3798	0.2465
0.4188	1.4690	12900	0.3822	0.2494
0.4424	1.5031	13200	0.3560	0.2449
0.4249	1.5373	13500	0.3630	0.2514
0.4287	1.5714	13800	0.3662	0.2417
0.3712	1.6056	14100	0.3714	0.2562
0.3893	1.6398	14400	0.3711	0.2333
0.3935	1.6739	14700	0.3715	0.2413
0.3982	1.7081	15000	0.3551	0.2482
0.4124	1.7422	15300	0.3519	0.2412
0.3853	1.7764	15600	0.3429	0.2418
0.4096	1.8106	15900	0.3407	0.2394
0.3816	1.8447	16200	0.3607	0.2370
0.3769	1.8789	16500	0.3601	0.2291
0.3428	1.9131	16800	0.3578	0.2283
0.3636	1.9472	17100	0.3485	0.2334
0.3594	1.9814	17400	0.3539	0.2341
0.3692	2.0155	17700	0.3383	0.2282
0.3295	2.0497	18000	0.3354	0.2374
0.3442	2.0839	18300	0.3393	0.2340
0.3306	2.1180	18600	0.3567	0.2382
0.3243	2.1522	18900	0.3410	0.2287
0.3426	2.1864	19200	0.3244	0.2323
0.3552	2.2205	19500	0.3356	0.2318
0.3558	2.2547	19800	0.3686	0.2225
0.3485	2.2888	20100	0.3485	0.2230
0.3195	2.3230	20400	0.3197	0.2230
0.3145	2.3572	20700	0.3312	0.2294
0.3238	2.3913	21000	0.3331	0.2210
0.3288	2.4255	21300	0.3172	0.2272
0.3398	2.4596	21600	0.3228	0.2182
0.3185	2.4940	21900	0.3057	0.2272
0.3152	2.5281	22200	0.3133	0.2175
0.312	2.5623	22500	0.3155	0.2155
0.3131	2.5965	22800	0.3087	0.2200
0.2993	2.6306	23100	0.3123	0.2216
0.2953	2.6648	23400	0.3116	0.2203
0.274	2.6989	23700	0.3221	0.2099
0.3043	2.7331	24000	0.3092	0.2131
0.2939	2.7673	24300	0.3084	0.2134
0.3063	2.8014	24600	0.3119	0.2094
0.3108	2.8356	24900	0.2987	0.2104
0.3188	2.8698	25200	0.3030	0.2082
0.2921	2.9039	25500	0.3051	0.2090
0.2994	2.9381	25800	0.2939	0.2148
0.2789	2.9722	26100	0.3012	0.2068
0.2902	3.0064	26400	0.2981	0.2138
0.2899	3.0406	26700	0.2931	0.2062
0.2796	3.0747	27000	0.2953	0.2067
0.287	3.1089	27300	0.3006	0.2105
0.2828	3.1431	27600	0.2916	0.2121
0.2798	3.1772	27900	0.2974	0.2060
0.2757	3.2114	28200	0.2908	0.2042
0.2694	3.2455	28500	0.2905	0.2058
0.262	3.2797	28800	0.2866	0.2048
0.2623	3.3139	29100	0.2794	0.2062
0.282	3.3480	29400	0.2814	0.2004
0.2655	3.3822	29700	0.2891	0.2006
0.2757	3.4163	30000	0.2845	0.1983
0.2686	3.4505	30300	0.2818	0.2013
0.2571	3.4847	30600	0.2825	0.2003
0.2681	3.5188	30900	0.2814	0.2051
0.2628	3.5530	31200	0.2831	0.1998
0.2625	3.5872	31500	0.2775	0.2032
0.2448	3.6213	31800	0.2770	0.1984
0.2599	3.6555	32100	0.2732	0.2002
0.2492	3.6896	32400	0.2880	0.1942
0.2666	3.7238	32700	0.2701	0.1984
0.257	3.7580	33000	0.2687	0.1997
0.2589	3.7921	33300	0.2665	0.1997
0.2735	3.8263	33600	0.2678	0.1990
0.2477	3.8604	33900	0.2704	0.1958
0.2525	3.8946	34200	0.2695	0.1946
0.2401	3.9288	34500	0.2732	0.1931
0.2585	3.9629	34800	0.2682	0.1945
0.2582	3.9971	35100	0.2688	0.1951

Framework versions

Transformers 4.41.1
Pytorch 2.1.2+cu121
Datasets 2.19.1
Tokenizers 0.19.1

Task: Automatic Speech Recognition
Library: transformers
Type: wav2vec2-bert
License: mit
Created: Jul 24, 2024
Updated: Aug 5, 2024