smolvlm-mmocr-sft-round-2

This model is a fine-tuned version of HuggingFaceTB/SmolVLM-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0196

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: 96
eval_batch_size: 96
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 768
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: 1000
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.5099	0.0054	20	0.4944
0.4426	0.0109	40	0.4381
0.4117	0.0163	60	0.3822
0.3378	0.0217	80	0.3352
0.3107	0.0271	100	0.2984
0.2839	0.0326	120	0.2696
0.2629	0.0380	140	0.2478
0.2467	0.0434	160	0.2305
0.2306	0.0489	180	0.2160
0.2283	0.0543	200	0.2037
0.1958	0.0597	220	0.1927
0.1887	0.0651	240	0.1830
0.1699	0.0706	260	0.1738
0.155	0.0760	280	0.1655
0.1579	0.0814	300	0.1575
0.1437	0.0869	320	0.1506
0.1549	0.0923	340	0.1437
0.1336	0.0977	360	0.1371
0.1268	0.1032	380	0.1314
0.139	0.1086	400	0.1258
0.1215	0.1140	420	0.1205
0.1103	0.1194	440	0.1156
0.1141	0.1249	460	0.1111
0.1175	0.1303	480	0.1066
0.0978	0.1357	500	0.1023
0.1016	0.1412	520	0.0986
0.0899	0.1466	540	0.0947
0.0952	0.1520	560	0.0909
0.0775	0.1574	580	0.0876
0.0862	0.1629	600	0.0845
0.0777	0.1683	620	0.0809
0.077	0.1737	640	0.0779
0.0757	0.1792	660	0.0752
0.0627	0.1846	680	0.0724
0.0635	0.1900	700	0.0701
0.0619	0.1954	720	0.0674
0.0593	0.2009	740	0.0651
0.0686	0.2063	760	0.0633
0.0626	0.2117	780	0.0613
0.0596	0.2172	800	0.0591
0.0447	0.2226	820	0.0575
0.0548	0.2280	840	0.0556
0.0555	0.2334	860	0.0546
0.0547	0.2389	880	0.0536
0.0461	0.2443	900	0.0512
0.0445	0.2497	920	0.0501
0.0386	0.2552	940	0.0491
0.0334	0.2606	960	0.0475
0.0498	0.2660	980	0.0458
0.0365	0.2715	1000	0.0452
0.0364	0.2769	1020	0.0442
0.0351	0.2823	1040	0.0433
0.0393	0.2878	1060	0.0418
0.0359	0.2932	1080	0.0413
0.0347	0.2986	1100	0.0405
0.0351	0.3040	1120	0.0397
0.0386	0.3095	1140	0.0385
0.04	0.3149	1160	0.0380
0.0288	0.3203	1180	0.0375
0.0279	0.3258	1200	0.0370
0.0285	0.3312	1220	0.0362
0.0308	0.3366	1240	0.0357
0.0302	0.3420	1260	0.0351
0.026	0.3475	1280	0.0349
0.0262	0.3529	1300	0.0345
0.0295	0.3583	1320	0.0341
0.0315	0.3638	1340	0.0335
0.0325	0.3692	1360	0.0327
0.0264	0.3746	1380	0.0324
0.0321	0.3800	1400	0.0319
0.0268	0.3855	1420	0.0312
0.0248	0.3909	1440	0.0313
0.0361	0.3963	1460	0.0312
0.0277	0.4018	1480	0.0303
0.0274	0.4072	1500	0.0300
0.028	0.4126	1520	0.0294
0.0223	0.4181	1540	0.0290
0.0247	0.4235	1560	0.0289
0.0265	0.4289	1580	0.0289
0.0285	0.4343	1600	0.0285
0.0284	0.4398	1620	0.0284
0.0255	0.4452	1640	0.0279
0.0246	0.4506	1660	0.0273
0.0196	0.4561	1680	0.0274
0.0272	0.4615	1700	0.0270
0.0206	0.4669	1720	0.0268
0.02	0.4723	1740	0.0268
0.0229	0.4778	1760	0.0268
0.0214	0.4832	1780	0.0260
0.0246	0.4886	1800	0.0261
0.0189	0.4941	1820	0.0254
0.0199	0.4995	1840	0.0253
0.0168	0.5049	1860	0.0251
0.0297	0.5103	1880	0.0250
0.0237	0.5158	1900	0.0245
0.0189	0.5212	1920	0.0250
0.0274	0.5266	1940	0.0243
0.0143	0.5321	1960	0.0243
0.0201	0.5375	1980	0.0241
0.0258	0.5429	2000	0.0241
0.0256	0.5484	2020	0.0239
0.018	0.5538	2040	0.0235
0.019	0.5592	2060	0.0236
0.0252	0.5646	2080	0.0233
0.0215	0.5701	2100	0.0234
0.0204	0.5755	2120	0.0235
0.0165	0.5809	2140	0.0230
0.018	0.5864	2160	0.0230
0.0236	0.5918	2180	0.0228
0.0199	0.5972	2200	0.0228
0.0153	0.6026	2220	0.0226
0.0166	0.6081	2240	0.0227
0.0212	0.6135	2260	0.0225
0.0213	0.6189	2280	0.0224
0.0165	0.6244	2300	0.0223
0.0197	0.6298	2320	0.0224
0.0219	0.6352	2340	0.0223
0.0277	0.6407	2360	0.0222
0.015	0.6461	2380	0.0220
0.0191	0.6515	2400	0.0218
0.0204	0.6569	2420	0.0220
0.0162	0.6624	2440	0.0219
0.0174	0.6678	2460	0.0219
0.0144	0.6732	2480	0.0217
0.0219	0.6787	2500	0.0215
0.0152	0.6841	2520	0.0217
0.0107	0.6895	2540	0.0215
0.0173	0.6949	2560	0.0213
0.0173	0.7004	2580	0.0214
0.0147	0.7058	2600	0.0214
0.015	0.7112	2620	0.0211
0.0158	0.7167	2640	0.0211
0.0128	0.7221	2660	0.0211
0.0185	0.7275	2680	0.0210
0.0214	0.7329	2700	0.0210
0.0162	0.7384	2720	0.0209
0.017	0.7438	2740	0.0210
0.0124	0.7492	2760	0.0211
0.0187	0.7547	2780	0.0207
0.0166	0.7601	2800	0.0209
0.0204	0.7655	2820	0.0207
0.0168	0.7710	2840	0.0207
0.0209	0.7764	2860	0.0206
0.0166	0.7818	2880	0.0206
0.0174	0.7872	2900	0.0207
0.0208	0.7927	2920	0.0205
0.013	0.7981	2940	0.0204
0.0112	0.8035	2960	0.0205
0.0174	0.8090	2980	0.0203
0.0169	0.8144	3000	0.0204
0.0129	0.8198	3020	0.0202
0.0191	0.8252	3040	0.0201
0.0166	0.8307	3060	0.0202
0.0182	0.8361	3080	0.0201
0.0208	0.8415	3100	0.0201
0.015	0.8470	3120	0.0200
0.0246	0.8524	3140	0.0200
0.0158	0.8578	3160	0.0201
0.0183	0.8633	3180	0.0201
0.0171	0.8687	3200	0.0201
0.0173	0.8741	3220	0.0199
0.017	0.8795	3240	0.0199
0.0123	0.8850	3260	0.0199
0.0144	0.8904	3280	0.0200
0.0149	0.8958	3300	0.0199
0.0139	0.9013	3320	0.0199
0.0199	0.9067	3340	0.0198
0.0164	0.9121	3360	0.0197
0.0154	0.9175	3380	0.0198
0.0225	0.9230	3400	0.0198
0.0232	0.9284	3420	0.0198
0.0135	0.9338	3440	0.0197
0.016	0.9393	3460	0.0197
0.0164	0.9447	3480	0.0197
0.0147	0.9501	3500	0.0197
0.015	0.9555	3520	0.0197
0.0215	0.9610	3540	0.0196
0.0174	0.9664	3560	0.0197
0.0203	0.9718	3580	0.0197
0.0191	0.9773	3600	0.0197
0.02	0.9827	3620	0.0196
0.0128	0.9881	3640	0.0196
0.0145	0.9936	3660	0.0197
0.0179	0.9990	3680	0.0196

Framework versions

PEFT 0.14.0
Transformers 4.49.0
Pytorch 2.5.1+cu124
Datasets 3.3.2
Tokenizers 0.21.1

chuuhtetnaing/smolvlm-mmocr-sft-round-2

smolvlm-mmocr-sft-round-2

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

为此模型做种