vit-base_rvl_tobacco

This model is a fine-tuned version of jordyvl/vit-base_rvl-cdip on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4152
• Accuracy: 0.905
• Brier Loss: 0.1584
• Nll: 0.7130
• F1 Micro: 0.905
• F1 Macro: 0.9056
• Ece: 0.1601
• Aurc: 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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	0.96	3	2.3234	0.045	0.9050	9.6090	0.045	0.0479	0.1570	0.9674
No log	1.96	6	2.3007	0.05	0.9005	8.5690	0.0500	0.0549	0.1567	0.9599
No log	2.96	9	2.2614	0.095	0.8924	6.9011	0.095	0.0853	0.1807	0.9128
No log	3.96	12	2.2062	0.255	0.8804	5.5442	0.255	0.1609	0.2738	0.7469
No log	4.96	15	2.1348	0.385	0.8636	4.0613	0.3850	0.2330	0.3605	0.4157
No log	5.96	18	2.0473	0.48	0.8410	2.5353	0.48	0.3152	0.4376	0.2329
No log	6.96	21	1.9483	0.64	0.8128	2.0469	0.64	0.5131	0.5355	0.1314
No log	7.96	24	1.8371	0.735	0.7783	1.7309	0.735	0.6333	0.5897	0.0802
No log	8.96	27	1.7227	0.775	0.7393	1.3371	0.775	0.6937	0.6049	0.0560
No log	9.96	30	1.6124	0.805	0.6978	1.1320	0.805	0.7319	0.5981	0.0462
No log	10.96	33	1.4990	0.82	0.6518	0.9973	0.82	0.7658	0.5882	0.0444
No log	11.96	36	1.3922	0.855	0.6064	0.8830	0.855	0.8127	0.5823	0.0397
No log	12.96	39	1.2985	0.865	0.5653	0.8957	0.865	0.8350	0.5604	0.0365
No log	13.96	42	1.2141	0.89	0.5271	0.6892	0.89	0.8733	0.5564	0.0331
No log	14.96	45	1.1402	0.895	0.4926	0.6695	0.895	0.8803	0.5341	0.0321
No log	15.96	48	1.0699	0.91	0.4596	0.6407	0.91	0.8999	0.5185	0.0285
No log	16.96	51	1.0037	0.91	0.4282	0.6163	0.91	0.8979	0.4831	0.0270
No log	17.96	54	0.9457	0.915	0.4004	0.6126	0.915	0.9011	0.4618	0.0247
No log	18.96	57	0.8914	0.915	0.3742	0.6066	0.915	0.9011	0.4426	0.0242
No log	19.96	60	0.8405	0.92	0.3495	0.5898	0.92	0.9102	0.4314	0.0216
No log	20.96	63	0.7995	0.915	0.3291	0.5934	0.915	0.9049	0.4033	0.0204
No log	21.96	66	0.7583	0.915	0.3089	0.5883	0.915	0.9049	0.3818	0.0206
No log	22.96	69	0.7228	0.915	0.2915	0.5835	0.915	0.9049	0.3707	0.0199
No log	23.96	72	0.6889	0.925	0.2747	0.5703	0.925	0.9169	0.3649	0.0191
No log	24.96	75	0.6624	0.925	0.2614	0.5769	0.925	0.9200	0.3375	0.0190
No log	25.96	78	0.6373	0.925	0.2491	0.5764	0.925	0.9218	0.3206	0.0191
No log	26.96	81	0.6106	0.93	0.2363	0.5570	0.93	0.9251	0.3276	0.0186
No log	27.96	84	0.5945	0.93	0.2281	0.5721	0.93	0.9251	0.3201	0.0187
No log	28.96	87	0.5780	0.92	0.2206	0.5668	0.92	0.9190	0.3008	0.0200
No log	29.96	90	0.5613	0.925	0.2125	0.5709	0.925	0.9218	0.2961	0.0191
No log	30.96	93	0.5456	0.925	0.2051	0.6155	0.925	0.9175	0.2764	0.0182
No log	31.96	96	0.5354	0.91	0.2008	0.6139	0.91	0.9104	0.2600	0.0187
No log	32.96	99	0.5248	0.91	0.1961	0.6078	0.91	0.9104	0.2610	0.0194
No log	33.96	102	0.5151	0.91	0.1915	0.6158	0.91	0.9084	0.2529	0.0186
No log	34.96	105	0.5066	0.91	0.1880	0.6121	0.91	0.9084	0.2409	0.0186
No log	35.96	108	0.4986	0.91	0.1846	0.6070	0.91	0.9084	0.2429	0.0186
No log	36.96	111	0.4920	0.91	0.1817	0.6208	0.91	0.9084	0.2380	0.0187
No log	37.96	114	0.4858	0.91	0.1793	0.6081	0.91	0.9084	0.2319	0.0185
No log	38.96	117	0.4792	0.91	0.1766	0.6044	0.91	0.9084	0.2276	0.0184
No log	39.96	120	0.4753	0.91	0.1749	0.6671	0.91	0.9084	0.2245	0.0185
No log	40.96	123	0.4704	0.905	0.1731	0.6137	0.905	0.9056	0.2321	0.0186
No log	41.96	126	0.4656	0.91	0.1714	0.6028	0.91	0.9084	0.2259	0.0187
No log	42.96	129	0.4624	0.91	0.1703	0.6048	0.91	0.9084	0.2080	0.0189
No log	43.96	132	0.4604	0.905	0.1695	0.6674	0.905	0.9056	0.2167	0.0187
No log	44.96	135	0.4553	0.905	0.1678	0.6190	0.905	0.9056	0.2130	0.0185
No log	45.96	138	0.4512	0.905	0.1663	0.6002	0.905	0.9056	0.2182	0.0186
No log	46.96	141	0.4513	0.905	0.1665	0.6681	0.905	0.9056	0.1902	0.0185
No log	47.96	144	0.4480	0.905	0.1656	0.6661	0.905	0.9056	0.1900	0.0186
No log	48.96	147	0.4451	0.905	0.1647	0.6085	0.905	0.9056	0.1969	0.0185
No log	49.96	150	0.4429	0.905	0.1638	0.6729	0.905	0.9056	0.1954	0.0186
No log	50.96	153	0.4416	0.905	0.1637	0.7300	0.905	0.9056	0.1730	0.0188
No log	51.96	156	0.4390	0.905	0.1627	0.6832	0.905	0.9056	0.1881	0.0187
No log	52.96	159	0.4377	0.905	0.1625	0.6708	0.905	0.9056	0.1724	0.0187
No log	53.96	162	0.4360	0.905	0.1620	0.7300	0.905	0.9056	0.1714	0.0189
No log	54.96	165	0.4338	0.905	0.1613	0.6734	0.905	0.9056	0.1923	0.0190
No log	55.96	168	0.4321	0.905	0.1609	0.6635	0.905	0.9056	0.1846	0.0189
No log	56.96	171	0.4326	0.905	0.1614	0.6722	0.905	0.9056	0.1851	0.0190
No log	57.96	174	0.4322	0.905	0.1613	0.7871	0.905	0.9056	0.1850	0.0191
No log	58.96	177	0.4286	0.905	0.1600	0.6660	0.905	0.9056	0.1733	0.0190
No log	59.96	180	0.4267	0.905	0.1596	0.6581	0.905	0.9056	0.1720	0.0190
No log	60.96	183	0.4277	0.905	0.1601	0.7252	0.905	0.9056	0.1772	0.0189
No log	61.96	186	0.4274	0.905	0.1601	0.7841	0.905	0.9056	0.1866	0.0192
No log	62.96	189	0.4264	0.905	0.1598	0.7830	0.905	0.9056	0.1669	0.0191
No log	63.96	192	0.4246	0.905	0.1595	0.7188	0.905	0.9056	0.1671	0.0191
No log	64.96	195	0.4236	0.905	0.1592	0.7170	0.905	0.9056	0.1762	0.0193
No log	65.96	198	0.4238	0.905	0.1594	0.7235	0.905	0.9056	0.1757	0.0192
No log	66.96	201	0.4227	0.905	0.1591	0.7218	0.905	0.9056	0.1724	0.0192
No log	67.96	204	0.4220	0.905	0.1590	0.7195	0.905	0.9056	0.1715	0.0191
No log	68.96	207	0.4214	0.905	0.1589	0.7201	0.905	0.9056	0.1708	0.0191
No log	69.96	210	0.4210	0.905	0.1588	0.7210	0.905	0.9056	0.1703	0.0193
No log	70.96	213	0.4211	0.905	0.1590	0.7226	0.905	0.9056	0.1697	0.0193
No log	71.96	216	0.4201	0.905	0.1587	0.7165	0.905	0.9056	0.1785	0.0193
No log	72.96	219	0.4194	0.905	0.1587	0.7145	0.905	0.9056	0.1780	0.0194
No log	73.96	222	0.4194	0.905	0.1587	0.7189	0.905	0.9056	0.1777	0.0194
No log	74.96	225	0.4192	0.905	0.1587	0.7193	0.905	0.9056	0.1770	0.0194
No log	75.96	228	0.4188	0.905	0.1586	0.7186	0.905	0.9056	0.1764	0.0192
No log	76.96	231	0.4180	0.905	0.1585	0.7148	0.905	0.9056	0.1786	0.0192
No log	77.96	234	0.4174	0.905	0.1584	0.7121	0.905	0.9056	0.1746	0.0193
No log	78.96	237	0.4178	0.905	0.1585	0.7159	0.905	0.9056	0.1720	0.0195
No log	79.96	240	0.4177	0.905	0.1586	0.7161	0.905	0.9056	0.1627	0.0195
No log	80.96	243	0.4173	0.905	0.1585	0.7147	0.905	0.9056	0.1627	0.0195
No log	81.96	246	0.4171	0.905	0.1585	0.7159	0.905	0.9056	0.1650	0.0195
No log	82.96	249	0.4162	0.905	0.1582	0.7135	0.905	0.9056	0.1742	0.0194
No log	83.96	252	0.4163	0.905	0.1584	0.7138	0.905	0.9056	0.1522	0.0196
No log	84.96	255	0.4161	0.905	0.1583	0.7136	0.905	0.9056	0.1616	0.0195
No log	85.96	258	0.4163	0.905	0.1585	0.7143	0.905	0.9056	0.1615	0.0196
No log	86.96	261	0.4161	0.905	0.1585	0.7132	0.905	0.9056	0.1614	0.0195
No log	87.96	264	0.4159	0.905	0.1584	0.7133	0.905	0.9056	0.1514	0.0195
No log	88.96	267	0.4157	0.905	0.1584	0.7132	0.905	0.9056	0.1513	0.0195
No log	89.96	270	0.4156	0.905	0.1584	0.7134	0.905	0.9056	0.1511	0.0195
No log	90.96	273	0.4153	0.905	0.1583	0.7124	0.905	0.9056	0.1605	0.0195
No log	91.96	276	0.4153	0.905	0.1584	0.7121	0.905	0.9056	0.1604	0.0195
No log	92.96	279	0.4154	0.905	0.1584	0.7127	0.905	0.9056	0.1603	0.0195
No log	93.96	282	0.4154	0.905	0.1585	0.7131	0.905	0.9056	0.1603	0.0195
No log	94.96	285	0.4154	0.905	0.1585	0.7132	0.905	0.9056	0.1603	0.0195
No log	95.96	288	0.4154	0.905	0.1585	0.7135	0.905	0.9056	0.1603	0.0196
No log	96.96	291	0.4153	0.905	0.1585	0.7133	0.905	0.9056	0.1602	0.0195
No log	97.96	294	0.4152	0.905	0.1584	0.7132	0.905	0.9056	0.1601	0.0196
No log	98.96	297	0.4152	0.905	0.1584	0.7130	0.905	0.9056	0.1601	0.0196
No log	99.96	300	0.4152	0.905	0.1584	0.7130	0.905	0.9056	0.1601	0.0196

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2