margin-element-detector-fm-resilient-puddle-10

This model is a fine-tuned version of microsoft/table-transformer-detection on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4052
• Loss Ce: 0.0393
• Loss Bbox: 0.0119
• Cardinality Error: 1.0210
• Giou: 84.6670

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
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Loss Ce	Loss Bbox	Cardinality Error	Giou
1.8005	0.5	1250	1.7181	0.3317	0.0619	1.8440	46.1650
1.6365	1.0	2500	1.5861	0.3064	0.0540	2.0670	49.5198
1.4739	1.5	3750	1.4081	0.2414	0.0487	1.2300	53.8370
1.3831	2.0	5000	1.2797	0.1926	0.0424	1.3180	56.2369
1.2362	2.5	6250	1.2517	0.1801	0.0406	1.3390	56.5658
1.2328	3.0	7500	1.2189	0.1650	0.0387	1.2300	56.9758
1.1675	3.5	8750	1.0386	0.1388	0.0317	1.1000	62.9430
1.1411	4.0	10000	1.0574	0.1392	0.0347	1.0590	62.7719
1.0822	4.5	11250	1.0113	0.1187	0.0337	1.0750	63.8054
1.0703	5.0	12500	0.9718	0.1181	0.0301	1.0770	64.8419
1.0278	5.5	13750	0.9538	0.1284	0.0276	1.1210	65.6340
1.044	6.0	15000	0.9157	0.1087	0.0294	1.0430	67.0038
0.9623	6.5	16250	0.9210	0.1135	0.0291	1.0630	66.9005
0.9883	7.0	17500	0.9465	0.1058	0.0311	1.0280	65.7425
0.953	7.5	18750	0.9267	0.0954	0.0292	1.0160	65.7261
0.9673	8.0	20000	0.8716	0.0904	0.0259	1.0230	67.4044
0.8954	8.5	21250	0.8415	0.0812	0.0256	1.0260	68.3924
0.9177	9.0	22500	0.8036	0.0819	0.0237	1.0170	69.8347
0.8572	9.5	23750	0.8165	0.0782	0.0234	1.0130	68.9332
0.8408	10.0	25000	0.8299	0.0767	0.0235	1.0390	68.2173
0.8281	10.5	26250	0.7925	0.0824	0.0229	1.0150	70.2080
0.8488	11.0	27500	0.8325	0.0718	0.0260	0.9950	68.4594
0.7916	11.5	28750	0.8020	0.0785	0.0231	1.0410	69.5891
0.8569	12.0	30000	0.7565	0.0681	0.0223	1.0180	71.1528
0.8023	12.5	31250	0.7649	0.0687	0.0217	1.0190	70.6185
0.776	13.0	32500	0.7613	0.0688	0.0237	0.9970	71.3041
0.7715	13.5	33750	0.7440	0.0689	0.0215	0.9850	71.6202
0.7823	14.0	35000	0.7766	0.0717	0.0220	1.0280	70.2445
0.7579	14.5	36250	0.7339	0.0613	0.0205	1.0510	71.4997
0.7693	15.0	37500	0.7738	0.0661	0.0225	1.0220	70.2403
0.713	15.5	38750	0.6801	0.0614	0.0190	1.0430	73.8128
0.6734	16.0	40000	0.7041	0.0623	0.0213	1.0100	73.2345
0.7289	16.5	41250	0.6959	0.0607	0.0209	1.0060	73.4663
0.7205	17.0	42500	0.7272	0.0704	0.0215	1.0110	72.5326
0.6855	17.5	43750	0.6586	0.0624	0.0195	1.0330	75.0753
0.6523	18.0	45000	0.6495	0.0557	0.0192	1.0380	75.1177
0.6519	18.5	46250	0.6763	0.0589	0.0198	1.0060	74.0859
0.6568	19.0	47500	0.6548	0.0758	0.0181	1.0200	75.5647
0.6254	19.5	48750	0.6494	0.0584	0.0193	1.0320	75.2703
0.6487	20.0	50000	0.6183	0.0624	0.0183	1.0570	76.7859
0.6287	20.5	51250	0.6432	0.0565	0.0193	1.0010	75.4949
0.6163	21.0	52500	0.6062	0.0485	0.0162	1.0110	76.1785
0.6029	21.5	53750	0.6158	0.0504	0.0174	1.0200	76.0916
0.622	22.0	55000	0.6186	0.0546	0.0180	0.9950	76.3034
0.597	22.5	56250	0.6172	0.0513	0.0180	1.0120	76.2164
0.5684	23.0	57500	0.5967	0.0527	0.0175	1.0250	77.1797
0.5899	23.5	58750	0.6035	0.0538	0.0178	1.0250	76.9589
0.5592	24.0	60000	0.6320	0.0548	0.0179	1.0180	75.6223
0.5994	24.5	61250	0.5444	0.0529	0.0159	1.0210	79.3936
0.5547	25.0	62500	0.5969	0.0527	0.0174	1.0320	77.1495
0.5135	25.5	63750	0.5651	0.0524	0.0163	1.0310	78.4524
0.5504	26.0	65000	0.5823	0.0451	0.0172	1.0150	77.4492
0.5342	26.5	66250	0.5905	0.0489	0.0169	1.0090	77.1484
0.5166	27.0	67500	0.5651	0.0488	0.0157	1.0010	78.1068
0.5311	27.5	68750	0.5585	0.0532	0.0162	1.0280	78.7836
0.5178	28.0	70000	0.5315	0.0451	0.0152	1.0190	79.4811
0.4967	28.5	71250	0.5399	0.0518	0.0151	1.0210	79.3648
0.5137	29.0	72500	0.5199	0.0461	0.0143	1.0310	79.8946
0.4903	29.5	73750	0.4885	0.0470	0.0144	1.0100	81.5240
0.4739	30.0	75000	0.4985	0.0447	0.0134	1.0150	80.6692
0.4455	30.5	76250	0.4999	0.0461	0.0140	1.0290	80.8051
0.4476	31.0	77500	0.4961	0.0466	0.0140	1.0090	81.0313
0.4581	31.5	78750	0.4980	0.0406	0.0141	1.0310	80.6620
0.4413	32.0	80000	0.5194	0.0431	0.0144	1.0300	79.7935
0.4332	32.5	81250	0.4861	0.0423	0.0139	1.0270	81.2911
0.444	33.0	82500	0.4515	0.0408	0.0127	1.0290	82.6487
0.4323	33.5	83750	0.4629	0.0434	0.0134	1.0300	82.3851
0.4299	34.0	85000	0.4602	0.0403	0.0129	1.0220	82.2341
0.403	34.5	86250	0.4693	0.0440	0.0133	1.0350	82.0647
0.4001	35.0	87500	0.4582	0.0397	0.0132	1.0210	82.3646
0.3987	35.5	88750	0.4354	0.0405	0.0125	1.0220	83.3753
0.3814	36.0	90000	0.4327	0.0397	0.0129	1.0290	83.5913
0.3694	36.5	91250	0.4285	0.0395	0.0128	1.0370	83.7543
0.3791	37.0	92500	0.4262	0.0382	0.0123	1.0200	83.6733
0.3646	37.5	93750	0.4133	0.0406	0.0123	1.0460	84.4284
0.3756	38.0	95000	0.4211	0.0397	0.0121	1.0080	83.9594
0.3566	38.5	96250	0.4125	0.0382	0.0120	1.0190	84.2887
0.3601	39.0	97500	0.4082	0.0395	0.0119	1.0320	84.5329
0.3483	39.5	98750	0.4064	0.0395	0.0119	1.0230	84.6185
0.3485	40.0	100000	0.4052	0.0393	0.0119	1.0210	84.6670

Framework versions

• Transformers 4.33.2
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.13.3