gg_mdl

This model is a fine-tuned version of openai/whisper-base on the gg_ds dataset. It achieves the following results on the evaluation set:

• Loss: 2.0181
• Cer: 26.4902

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: 4
• eval_batch_size: 4
• seed: 42
• 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: 500
• training_steps: 300000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.0028	0.8	1000	1.5737	27.8905
0.0059	1.6	2000	1.5694	27.5053
0.0039	2.4	3000	1.5481	27.8554
0.0042	3.2	4000	1.5456	27.6774
0.0029	4.0	5000	1.5568	26.8684
0.0072	4.8	6000	1.5703	27.7266
0.0026	5.6	7000	1.5721	30.2872
0.002	6.4	8000	1.5597	27.0569
0.0077	7.2	9000	1.5680	27.3742
0.0059	8.0	10000	1.5891	28.6492
0.0023	8.8	11000	1.5614	26.9854
0.0043	9.6	12000	1.5679	26.9480
0.0027	10.4	13000	1.5807	26.9257
0.0024	11.2	14000	1.5903	26.1460
0.0028	12.0	15000	1.5812	30.4265
0.0043	12.8	16000	1.5962	28.9115
0.0019	13.6	17000	1.5915	28.0895
0.0066	14.4	18000	1.6038	27.1283
0.0052	15.2	19000	1.6088	27.0370
0.0015	16.0	20000	1.6055	26.7782
0.0076	16.8	21000	1.6063	26.5745
0.0074	17.6	22000	1.5864	26.1214
0.0009	18.4	23000	1.6042	28.1539
0.0025	19.2	24000	1.6047	27.0206
0.0041	20.0	25000	1.6130	26.9737
0.0031	20.8	26000	1.6087	28.4068
0.0014	21.6	27000	1.5997	27.6739
0.0003	22.4	28000	1.6045	26.3309
0.006	23.2	29000	1.6061	27.3074
0.0031	24.0	30000	1.6221	29.8739
0.0015	24.8	31000	1.6350	27.4784
0.0028	25.6	32000	1.6271	26.8274
0.0015	26.4	33000	1.6386	28.1504
0.0007	27.2	34000	1.6262	26.5054
0.0035	28.0	35000	1.6441	30.8761
0.0029	28.8	36000	1.6650	27.3988
0.0014	29.6	37000	1.6366	27.2957
0.0051	30.4	38000	1.6433	26.0230
0.0007	31.2	39000	1.6476	27.1224
0.0042	32.0	40000	1.6526	27.3367
0.001	32.8	41000	1.6606	26.5113
0.0005	33.6	42000	1.6455	28.5239
0.0064	34.4	43000	1.6537	28.2324
0.0005	35.2	44000	1.6589	26.0980
0.0035	36.0	45000	1.6617	26.6412
0.0014	36.8	46000	1.6698	27.2606
0.0007	37.6	47000	1.6751	26.9726
0.0036	38.4	48000	1.6790	26.9620
0.0049	39.2	49000	1.6914	26.9222
0.002	40.0	50000	1.7004	27.1728
0.0032	40.8	51000	1.7019	26.6096
0.0012	41.6	52000	1.7076	27.5006
0.0004	42.4	53000	1.7054	26.6553
0.0029	43.2	54000	1.6880	27.3952
0.0013	44.0	55000	1.6983	27.7722
0.0021	44.8	56000	1.7000	28.0275
0.0007	45.6	57000	1.6831	27.9268
0.0007	46.4	58000	1.6989	26.1284
0.0025	47.2	59000	1.6858	27.2372
0.0003	48.0	60000	1.7004	29.3213
0.0011	48.8	61000	1.7122	26.9433
0.0017	49.6	62000	1.7014	25.8474
0.0023	50.4	63000	1.7186	26.5546
0.0031	51.2	64000	1.6963	27.2735
0.0023	52.0	65000	1.7043	26.9117
0.0004	52.8	66000	1.7068	26.3064
0.0003	53.6	67000	1.7196	27.1517
0.0012	54.4	68000	1.7213	26.9737
0.0013	55.2	69000	1.7114	26.5347
0.0007	56.0	70000	1.7245	28.1001
0.0002	56.8	71000	1.7138	27.8788
0.0001	57.6	72000	1.7182	27.3543
0.0009	58.4	73000	1.7354	28.0942
0.0023	59.2	74000	1.7314	27.0077
0.0013	60.0	75000	1.7350	28.0439
0.0019	60.8	76000	1.7322	27.7512
0.0025	61.6	77000	1.7620	27.5978
0.0005	62.4	78000	1.7201	27.7102
0.0043	63.2	79000	1.7405	27.8565
0.0006	64.0	80000	1.7550	29.4407
0.0004	64.8	81000	1.7410	27.6189
0.0002	65.6	82000	1.7312	29.3189
0.0006	66.4	83000	1.7476	26.5312
0.0033	67.2	84000	1.7571	26.8098
0.0041	68.0	85000	1.7463	27.2512
0.0014	68.8	86000	1.7401	26.3485
0.0001	69.6	87000	1.7465	27.6622
0.0011	70.4	88000	1.7494	26.9468
0.0001	71.2	89000	1.7423	28.5684
0.0037	72.0	90000	1.7728	29.7896
0.0002	72.8	91000	1.7555	26.9140
0.0006	73.6	92000	1.7685	26.7606
0.0014	74.4	93000	1.7494	26.8204
0.001	75.2	94000	1.7719	26.7150
0.0005	76.0	95000	1.7754	26.6892
0.0022	76.8	96000	1.7698	27.4807
0.0017	77.6	97000	1.7830	27.7465
0.0001	78.4	98000	1.7751	27.0487
0.0025	79.2	99000	1.7768	27.1505
0.0001	80.0	100000	1.7671	27.2805
0.0019	80.8	101000	1.7910	27.3027
0.0031	81.6	102000	1.7965	27.6809
0.0016	82.4	103000	1.7893	28.0146
0.0002	83.2	104000	1.7939	26.7384
0.0001	84.0	105000	1.7925	27.3156
0.0001	84.8	106000	1.7866	27.8718
0.0001	85.6	107000	1.7789	27.0171
0.0001	86.4	108000	1.7738	26.3977
0.0005	87.2	109000	1.7748	27.8577
0.0015	88.0	110000	1.7922	26.6611
0.0003	88.8	111000	1.7987	28.0486
0.0017	89.6	112000	1.7901	28.0860
0.0001	90.4	113000	1.8013	27.7523
0.0001	91.2	114000	1.8045	26.9796
0.0005	92.0	115000	1.7989	27.0112
0.0024	92.8	116000	1.8068	26.9597
0.0001	93.6	117000	1.8033	29.2651
0.0001	94.4	118000	1.7955	28.6984
0.0004	95.2	119000	1.7956	27.0920
0.0005	96.0	120000	1.7868	27.3426
0.0011	96.8	121000	1.8209	26.8005
0.0005	97.6	122000	1.8152	29.9816
0.002	98.4	123000	1.8174	26.7255
0.0001	99.2	124000	1.8194	26.9164
0.0004	100.0	125000	1.8307	27.7289
0.0001	100.8	126000	1.8151	26.9609
0.0001	101.6	127000	1.8080	28.0158
0.0001	102.4	128000	1.8349	26.7571
0.0002	103.2	129000	1.8371	27.0686
0.0006	104.0	130000	1.8133	27.4842
0.0001	104.8	131000	1.8246	26.5768
0.0005	105.6	132000	1.8180	26.7489
0.0001	106.4	133000	1.8261	27.4409
0.0001	107.2	134000	1.8101	26.7443
0.0003	108.0	135000	1.8164	27.3800
0.0001	108.8	136000	1.8152	26.9679
0.0001	109.6	137000	1.8121	26.6775
0.0001	110.4	138000	1.8317	27.7968
0.0	111.2	139000	1.8266	26.6869
0.0001	112.0	140000	1.8331	27.2067
0.0045	112.8	141000	1.8353	27.0276
0.0024	113.6	142000	1.8416	28.0345
0.0005	114.4	143000	1.8359	27.8460
0.0	115.2	144000	1.8390	27.1892
0.0	116.0	145000	1.8311	27.2934
0.0	116.8	146000	1.8473	27.4386
0.0016	117.6	147000	1.8554	27.3964
0.0001	118.4	148000	1.8608	26.5148
0.0021	119.2	149000	1.8582	26.9058
0.0003	120.0	150000	1.8574	26.9269
0.0002	120.8	151000	1.8568	27.2079
0.0	121.6	152000	1.8623	26.5417
0.0001	122.4	153000	1.8500	27.2009
0.0	123.2	154000	1.8604	27.6236
0.0	124.0	155000	1.8739	27.8203
0.0001	124.8	156000	1.8705	26.8215
0.0	125.6	157000	1.8521	27.1283
0.0	126.4	158000	1.8607	26.5241
0.0	127.2	159000	1.8646	27.1423
0.0001	128.0	160000	1.8665	26.9538
0.0002	128.8	161000	1.8768	26.7841
0.0002	129.6	162000	1.8722	26.7864
0.0005	130.4	163000	1.8626	27.3695
0.0	131.2	164000	1.8646	27.8425
0.0	132.0	165000	1.8758	27.8062
0.0002	132.8	166000	1.8780	28.5110
0.0	133.6	167000	1.8672	26.1319
0.0	134.4	168000	1.8833	26.4422
0.0009	135.2	169000	1.8828	27.4421
0.0	136.0	170000	1.8933	27.2243
0.0001	136.8	171000	1.8913	27.2302
0.0	137.6	172000	1.8941	27.2746
0.0001	138.4	173000	1.8873	26.5089
0.0004	139.2	174000	1.8966	26.7969
0.0	140.0	175000	1.8916	26.6611
0.0	140.8	176000	1.8890	26.4199
0.0	141.6	177000	1.8991	28.7066
0.0	142.4	178000	1.8963	27.2021
0.0	143.2	179000	1.8996	27.7231
0.0001	144.0	180000	1.9000	28.4513
0.0	144.8	181000	1.9029	27.0428
0.0	145.6	182000	1.9119	27.1540
0.0	146.4	183000	1.8947	26.8684
0.0	147.2	184000	1.9096	27.1131
0.0	148.0	185000	1.9065	25.9961
0.0	148.8	186000	1.9112	27.3004
0.0013	149.6	187000	1.9016	27.0182
0.0	150.4	188000	1.9075	26.8637
0.0	151.2	189000	1.9189	27.3016
0.0	152.0	190000	1.9179	28.9431
0.0	152.8	191000	1.9277	27.1283
0.0	153.6	192000	1.9123	27.5463
0.0001	154.4	193000	1.9066	26.6459
0.0002	155.2	194000	1.9222	26.8168
0.0	156.0	195000	1.9263	27.1435
0.0	156.8	196000	1.9363	26.9187
0.0	157.6	197000	1.9299	26.0546
0.0	158.4	198000	1.9429	27.1704
0.0014	159.2	199000	1.9413	26.4609
0.0	160.0	200000	1.9294	26.8567
0.0	160.8	201000	1.9351	27.6727
0.0	161.6	202000	1.9396	26.8297
0.0	162.4	203000	1.9388	26.9292
0.0	163.2	204000	1.9436	26.8531
0.0	164.0	205000	1.9439	27.5486
0.0	164.8	206000	1.9380	27.5252
0.0	165.6	207000	1.9396	26.4843
0.0	166.4	208000	1.9379	26.1846
0.0011	167.2	209000	1.9598	27.2407
0.0	168.0	210000	1.9474	26.6834
0.0	168.8	211000	1.9509	27.3367
0.0	169.6	212000	1.9567	27.4948
0.0	170.4	213000	1.9584	27.3671
0.0009	171.2	214000	1.9578	26.8168
0.0	172.0	215000	1.9477	27.9362
0.0007	172.8	216000	1.9651	27.3484
0.0	173.6	217000	1.9491	26.4515
0.0	174.4	218000	1.9434	27.3507
0.0001	175.2	219000	1.9572	27.3133
0.0	176.0	220000	1.9570	27.3812
0.0	176.8	221000	1.9577	27.4339
0.0	177.6	222000	1.9655	27.4924
0.0	178.4	223000	1.9625	27.2021
0.0	179.2	224000	1.9601	27.0346
0.0	180.0	225000	1.9703	26.8988
0.0	180.8	226000	1.9747	26.4539
0.0	181.6	227000	1.9728	26.4106
0.0	182.4	228000	1.9776	27.2372
0.0	183.2	229000	1.9866	26.7969
0.0	184.0	230000	1.9857	26.9164
0.0	184.8	231000	1.9847	26.5113
0.0	185.6	232000	1.9850	27.0897
0.0004	186.4	233000	1.9967	27.4749
0.0	187.2	234000	1.9906	26.5464
0.0	188.0	235000	2.0016	27.7336
0.0	188.8	236000	2.0036	26.6775
0.0	189.6	237000	1.9978	26.6119
0.0	190.4	238000	1.9968	27.2711
0.0	191.2	239000	1.9970	26.6319
0.0	192.0	240000	1.9969	26.8812
0.0	192.8	241000	2.0076	27.1201
0.0	193.6	242000	2.0073	26.4644
0.0	194.4	243000	2.0097	26.0371
0.0	195.2	244000	2.0108	25.8544
0.0	196.0	245000	2.0138	26.1998
0.0	196.8	246000	2.0177	26.4761
0.0	197.6	247000	2.0226	26.4925
0.0	198.4	248000	2.0277	27.0194
0.0	199.2	249000	2.0331	26.4059
0.0	200.0	250000	2.0388	26.0980
0.0	200.8	251000	2.0451	26.4562
0.0	201.6	252000	2.0527	26.1085
0.0	202.4	253000	2.0578	26.6529
0.0	203.2	254000	2.0631	26.5195
0.0	204.0	255000	2.0665	26.2314
0.0	204.8	256000	2.0711	26.5312
0.0	205.6	257000	2.0749	26.7021
0.0	206.4	258000	2.0747	26.2525
0.0	207.2	259000	2.0755	26.0277
0.0	208.0	260000	2.0746	25.9457
0.0	208.8	261000	2.0739	25.8404
0.0	209.6	262000	2.0720	25.7151
0.0	210.4	263000	2.0695	25.7151
0.0	211.2	264000	2.0669	25.6156
0.0	212.0	265000	2.0652	25.9130
0.0	212.8	266000	2.0625	25.7795
0.0	213.6	267000	2.0598	26.0827
0.0	214.4	268000	2.0576	25.6928
0.0	215.2	269000	2.0546	26.0628
0.0	216.0	270000	2.0530	25.6472
0.0	216.8	271000	2.0506	25.8076
0.0	217.6	272000	2.0476	25.9200
0.0	218.4	273000	2.0452	26.2595
0.0	219.2	274000	2.0437	26.0816
0.0	220.0	275000	2.0422	26.0382
0.0	220.8	276000	2.0401	26.0078
0.0	221.6	277000	2.0380	26.5440
0.0	222.4	278000	2.0358	26.1401
0.0	223.2	279000	2.0347	26.5487
0.0	224.0	280000	2.0334	26.1623
0.0	224.8	281000	2.0321	26.3743
0.0	225.6	282000	2.0303	26.2630
0.0	226.4	283000	2.0290	26.5604
0.0	227.2	284000	2.0280	26.7618
0.0	228.0	285000	2.0269	26.8859
0.0	228.8	286000	2.0256	26.7279
0.0	229.6	287000	2.0243	26.3871
0.0	230.4	288000	2.0238	26.6049
0.0	231.2	289000	2.0223	26.5452
0.0	232.0	290000	2.0222	26.4761
0.0	232.8	291000	2.0215	26.3497
0.0	233.6	292000	2.0206	26.4024
0.0	234.4	293000	2.0202	26.7899
0.0	235.2	294000	2.0196	26.8051
0.0	236.0	295000	2.0192	26.7466
0.0	236.8	296000	2.0187	26.5686
0.0	237.6	297000	2.0185	26.5956
0.0	238.4	298000	2.0183	26.2747
0.0	239.2	299000	2.0182	26.5253
0.0	240.0	300000	2.0181	26.4902

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.4.1
• Datasets 3.1.0
• Tokenizers 0.20.1