Modelpublic

paulo037/Qwen3-VL-2B-Instruct-legal-extraction

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face

0likes30downloads

Model Card

Qwen3-VL-2B-Instruct-legal-extraction

This model is a fine-tuned version of Qwen/Qwen3-VL-2B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 0.3722

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: 0.0001
—trainbatchsize: 1
—evalbatchsize: 1
—seed: 14
—gradientaccumulationsteps: 8
—totaltrainbatch_size: 8
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: linear
—lrschedulerwarmup_ratio: 0.03
—num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.5737	0.0101	25	1.3491
1.1111	0.0202	50	0.8806
0.6967	0.0303	75	0.6069
0.5292	0.0404	100	0.5561
0.495	0.0505	125	0.5374
0.4728	0.0606	150	0.5228
0.521	0.0707	175	0.5121
0.5131	0.0808	200	0.5018
0.5235	0.0909	225	0.4929
0.4913	0.1010	250	0.4870
0.4508	0.1111	275	0.4791
0.4849	0.1212	300	0.4731
0.4293	0.1313	325	0.4682
0.4145	0.1414	350	0.4656
0.4397	0.1515	375	0.4604
0.4565	0.1616	400	0.4570
0.461	0.1717	425	0.4585
0.4165	0.1818	450	0.4537
0.4327	0.1919	475	0.4483
0.4475	0.2020	500	0.4456
0.44	0.2121	525	0.4426
0.4106	0.2222	550	0.4413
0.4052	0.2323	575	0.4378
0.3968	0.2424	600	0.4352
0.3984	0.2525	625	0.4330
0.3773	0.2626	650	0.4315
0.4006	0.2727	675	0.4298
0.3933	0.2828	700	0.4265
0.4247	0.2929	725	0.4251
0.3744	0.3030	750	0.4226
0.3718	0.3131	775	0.4222
0.355	0.3232	800	0.4210
0.3746	0.3334	825	0.4186
0.4148	0.3435	850	0.4166
0.3809	0.3536	875	0.4158
0.394	0.3637	900	0.4139
0.4024	0.3738	925	0.4138
0.4156	0.3839	950	0.4126
0.3995	0.3940	975	0.4102
0.396	0.4041	1000	0.4091
0.3549	0.4142	1025	0.4082
0.3818	0.4243	1050	0.4078
0.4186	0.4344	1075	0.4065
0.4159	0.4445	1100	0.4050
0.4034	0.4546	1125	0.4044
0.3854	0.4647	1150	0.4029
0.3874	0.4748	1175	0.4021
0.4151	0.4849	1200	0.4016
0.3686	0.4950	1225	0.3994
0.3729	0.5051	1250	0.3979
0.342	0.5152	1275	0.3977
0.3834	0.5253	1300	0.3966
0.3774	0.5354	1325	0.3957
0.3874	0.5455	1350	0.3942
0.374	0.5556	1375	0.3929
0.3878	0.5657	1400	0.3920
0.3214	0.5758	1425	0.3915
0.3547	0.5859	1450	0.3904
0.3421	0.5960	1475	0.3900
0.3633	0.6061	1500	0.3894
0.3714	0.6162	1525	0.3889
0.3354	0.6263	1550	0.3882
0.3896	0.6364	1575	0.3887
0.3625	0.6465	1600	0.3864
0.378	0.6566	1625	0.3856
0.3311	0.6667	1650	0.3849
0.4433	0.6768	1675	0.3840
0.3542	0.6869	1700	0.3835
0.3603	0.6970	1725	0.3830
0.3829	0.7071	1750	0.3822
0.3577	0.7172	1775	0.3824
0.3956	0.7273	1800	0.3821
0.3698	0.7374	1825	0.3812
0.3628	0.7475	1850	0.3807
0.3791	0.7576	1875	0.3807
0.3785	0.7677	1900	0.3798
0.3853	0.7778	1925	0.3791
0.3823	0.7879	1950	0.3787
0.3516	0.7980	1975	0.3780
0.373	0.8081	2000	0.3776
0.3629	0.8182	2025	0.3771
0.3567	0.8283	2050	0.3766
0.3675	0.8384	2075	0.3760
0.3652	0.8485	2100	0.3752
0.359	0.8586	2125	0.3752
0.334	0.8687	2150	0.3746
0.3504	0.8788	2175	0.3748
0.3782	0.8889	2200	0.3743
0.379	0.8990	2225	0.3744
0.358	0.9091	2250	0.3740
0.3665	0.9192	2275	0.3737
0.3702	0.9293	2300	0.3734
0.3303	0.9394	2325	0.3732
0.3493	0.9495	2350	0.3729
0.3655	0.9596	2375	0.3728
0.3417	0.9697	2400	0.3725
0.3497	0.9798	2425	0.3723
0.3416	0.9899	2450	0.3722
0.378	1.0	2475	0.3722

Framework versions

—PEFT 0.17.1
—Transformers 4.57.0
—Pytorch 2.11.0a0+eb65b36914.nv26.02
—Datasets 4.1.1
—Tokenizers 0.22.2