dit-finetuned_rvl_tobacco_crl_allv2

This model is a fine-tuned version of microsoft/dit-base-finetuned-rvlcdip on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5457
• Accuracy: 0.93
• Brier Loss: 0.1595
• Nll: 0.7692
• F1 Micro: 0.93
• F1 Macro: 0.9154
• Ece: 0.2119
• Aurc: 0.0106

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	2.3428	0.005	0.9059	8.8950	0.005	0.0048	0.1391	0.9945
No log	1.96	6	2.3437	0.005	0.9052	8.8527	0.005	0.0048	0.1390	0.9942
No log	2.96	9	2.3397	0.005	0.9038	8.6108	0.005	0.0048	0.1389	0.9940
No log	3.96	12	2.3292	0.01	0.9018	7.9364	0.01	0.0091	0.1413	0.9925
No log	4.96	15	2.3200	0.02	0.8991	7.1029	0.02	0.0303	0.1462	0.9876
No log	5.96	18	2.3074	0.05	0.8958	6.6583	0.0500	0.0527	0.1638	0.9775
No log	6.96	21	2.2903	0.155	0.8918	6.3637	0.155	0.1012	0.2321	0.9477
No log	7.96	24	2.2706	0.18	0.8869	6.0129	0.18	0.1088	0.2487	0.9279
No log	8.96	27	2.2479	0.195	0.8810	4.5304	0.195	0.1154	0.2509	0.8991
No log	9.96	30	2.2212	0.215	0.8734	3.2783	0.2150	0.1365	0.2504	0.8008
No log	10.96	33	2.1853	0.305	0.8631	2.6641	0.305	0.1889	0.3073	0.6426
No log	11.96	36	2.1391	0.405	0.8489	2.2824	0.405	0.2593	0.3747	0.4193
No log	12.96	39	2.0878	0.485	0.8327	1.8576	0.485	0.3408	0.4203	0.2962
No log	13.96	42	2.0311	0.58	0.8157	1.4206	0.58	0.4327	0.4908	0.2186
No log	14.96	45	1.9707	0.605	0.7970	1.2940	0.605	0.4601	0.4866	0.1784
No log	15.96	48	1.9196	0.64	0.7803	1.2372	0.64	0.4958	0.5098	0.1483
No log	16.96	51	1.8690	0.68	0.7630	1.1960	0.68	0.5548	0.5403	0.1286
No log	17.96	54	1.8192	0.73	0.7448	1.1405	0.7300	0.6227	0.5655	0.0954
No log	18.96	57	1.7691	0.8	0.7272	1.1234	0.8000	0.6965	0.6131	0.0618
No log	19.96	60	1.7210	0.835	0.7096	1.0912	0.835	0.7380	0.6351	0.0463
No log	20.96	63	1.6734	0.84	0.6912	1.0707	0.8400	0.7410	0.6305	0.0404
No log	21.96	66	1.6273	0.85	0.6724	1.0024	0.85	0.7458	0.6316	0.0335
No log	22.96	69	1.5782	0.86	0.6525	0.9764	0.8600	0.7612	0.6265	0.0289
No log	23.96	72	1.5293	0.87	0.6327	0.9580	0.87	0.7690	0.6211	0.0227
No log	24.96	75	1.4840	0.87	0.6131	0.9549	0.87	0.7730	0.6157	0.0223
No log	25.96	78	1.4425	0.88	0.5937	0.9521	0.88	0.7829	0.6121	0.0209
No log	26.96	81	1.4013	0.875	0.5740	0.9456	0.875	0.7799	0.5931	0.0216
No log	27.96	84	1.3626	0.875	0.5550	0.9392	0.875	0.7828	0.5840	0.0219
No log	28.96	87	1.3237	0.87	0.5360	0.9386	0.87	0.7802	0.5656	0.0225
No log	29.96	90	1.2847	0.875	0.5166	0.9371	0.875	0.7854	0.5605	0.0208
No log	30.96	93	1.2515	0.88	0.4993	0.8803	0.88	0.7875	0.5431	0.0205
No log	31.96	96	1.2196	0.89	0.4824	0.9276	0.89	0.8120	0.5328	0.0192
No log	32.96	99	1.1854	0.885	0.4648	0.9279	0.885	0.8020	0.5183	0.0195
No log	33.96	102	1.1523	0.895	0.4480	0.9823	0.895	0.8347	0.5018	0.0189
No log	34.96	105	1.1231	0.895	0.4324	0.9680	0.895	0.8347	0.4931	0.0187
No log	35.96	108	1.0944	0.905	0.4174	0.9642	0.905	0.8527	0.4880	0.0173
No log	36.96	111	1.0648	0.91	0.4020	0.9006	0.91	0.8720	0.4702	0.0183
No log	37.96	114	1.0380	0.925	0.3885	0.8823	0.925	0.9005	0.4695	0.0160
No log	38.96	117	1.0127	0.92	0.3758	0.8700	0.92	0.8954	0.4523	0.0171
No log	39.96	120	0.9882	0.915	0.3629	0.8668	0.915	0.8864	0.4399	0.0170
No log	40.96	123	0.9655	0.93	0.3509	0.8642	0.93	0.9048	0.4435	0.0150
No log	41.96	126	0.9452	0.925	0.3412	0.8563	0.925	0.9016	0.4240	0.0159
No log	42.96	129	0.9248	0.925	0.3312	0.8581	0.925	0.9016	0.4130	0.0160
No log	43.96	132	0.9037	0.935	0.3207	0.8536	0.935	0.9204	0.4066	0.0159
No log	44.96	135	0.8859	0.93	0.3120	0.8420	0.93	0.9154	0.3929	0.0166
No log	45.96	138	0.8695	0.93	0.3039	0.8342	0.93	0.9154	0.3947	0.0163
No log	46.96	141	0.8536	0.935	0.2959	0.8340	0.935	0.9204	0.3970	0.0152
No log	47.96	144	0.8381	0.935	0.2889	0.8297	0.935	0.9204	0.3877	0.0150
No log	48.96	147	0.8217	0.94	0.2810	0.8237	0.94	0.9269	0.3783	0.0147
No log	49.96	150	0.8060	0.94	0.2725	0.8241	0.94	0.9277	0.3762	0.0142
No log	50.96	153	0.7892	0.935	0.2642	0.8185	0.935	0.9245	0.3522	0.0139
No log	51.96	156	0.7750	0.94	0.2577	0.8128	0.94	0.9330	0.3512	0.0131
No log	52.96	159	0.7602	0.94	0.2517	0.8020	0.94	0.9330	0.3517	0.0135
No log	53.96	162	0.7457	0.94	0.2449	0.7927	0.94	0.9330	0.3443	0.0134
No log	54.96	165	0.7342	0.94	0.2393	0.8457	0.94	0.9330	0.3248	0.0130
No log	55.96	168	0.7235	0.94	0.2344	0.8500	0.94	0.9330	0.3244	0.0127
No log	56.96	171	0.7161	0.935	0.2303	0.8536	0.935	0.9214	0.3181	0.0129
No log	57.96	174	0.7052	0.935	0.2251	0.8537	0.935	0.9214	0.3122	0.0128
No log	58.96	177	0.6930	0.935	0.2192	0.8442	0.935	0.9214	0.3123	0.0121
No log	59.96	180	0.6830	0.94	0.2146	0.8339	0.94	0.9263	0.3003	0.0119
No log	60.96	183	0.6735	0.94	0.2108	0.8266	0.94	0.9263	0.2944	0.0120
No log	61.96	186	0.6654	0.935	0.2068	0.8249	0.935	0.9231	0.3001	0.0120
No log	62.96	189	0.6572	0.935	0.2029	0.8228	0.935	0.9231	0.2784	0.0115
No log	63.96	192	0.6526	0.935	0.2001	0.8257	0.935	0.9231	0.2749	0.0116
No log	64.96	195	0.6448	0.935	0.1977	0.8244	0.935	0.9244	0.2643	0.0118
No log	65.96	198	0.6366	0.935	0.1944	0.8169	0.935	0.9244	0.2607	0.0115
No log	66.96	201	0.6281	0.935	0.1908	0.8088	0.935	0.9231	0.2745	0.0113
No log	67.96	204	0.6206	0.935	0.1876	0.8037	0.935	0.9231	0.2784	0.0112
No log	68.96	207	0.6143	0.935	0.1853	0.8025	0.935	0.9231	0.2662	0.0112
No log	69.96	210	0.6102	0.94	0.1839	0.8010	0.94	0.9330	0.2584	0.0113
No log	70.96	213	0.6053	0.94	0.1822	0.7991	0.94	0.9330	0.2639	0.0113
No log	71.96	216	0.6000	0.94	0.1802	0.7949	0.94	0.9330	0.2516	0.0113
No log	72.96	219	0.5941	0.94	0.1773	0.7891	0.94	0.9330	0.2670	0.0112
No log	73.96	222	0.5888	0.94	0.1747	0.7846	0.94	0.9330	0.2457	0.0109
No log	74.96	225	0.5837	0.94	0.1724	0.7831	0.94	0.9330	0.2425	0.0107
No log	75.96	228	0.5801	0.94	0.1711	0.7857	0.94	0.9330	0.2411	0.0108
No log	76.96	231	0.5783	0.935	0.1708	0.7884	0.935	0.9231	0.2342	0.0110
No log	77.96	234	0.5767	0.93	0.1707	0.7882	0.93	0.9139	0.2375	0.0109
No log	78.96	237	0.5754	0.93	0.1703	0.7863	0.93	0.9154	0.2255	0.0109
No log	79.96	240	0.5732	0.93	0.1694	0.7848	0.93	0.9154	0.2365	0.0111
No log	80.96	243	0.5715	0.93	0.1685	0.7834	0.93	0.9154	0.2433	0.0111
No log	81.96	246	0.5675	0.93	0.1672	0.7803	0.93	0.9154	0.2423	0.0110
No log	82.96	249	0.5648	0.93	0.1661	0.7773	0.93	0.9154	0.2352	0.0108
No log	83.96	252	0.5624	0.93	0.1653	0.7753	0.93	0.9154	0.2343	0.0108
No log	84.96	255	0.5608	0.93	0.1646	0.7746	0.93	0.9154	0.2332	0.0108
No log	85.96	258	0.5589	0.93	0.1642	0.7748	0.93	0.9154	0.2310	0.0110
No log	86.96	261	0.5574	0.93	0.1638	0.7741	0.93	0.9154	0.2286	0.0111
No log	87.96	264	0.5559	0.93	0.1636	0.7739	0.93	0.9154	0.2274	0.0108
No log	88.96	267	0.5555	0.93	0.1631	0.7748	0.93	0.9154	0.2264	0.0109
No log	89.96	270	0.5539	0.93	0.1623	0.7743	0.93	0.9154	0.2255	0.0108
No log	90.96	273	0.5525	0.93	0.1616	0.7738	0.93	0.9154	0.2251	0.0108
No log	91.96	276	0.5508	0.93	0.1612	0.7726	0.93	0.9154	0.2254	0.0108
No log	92.96	279	0.5498	0.93	0.1610	0.7721	0.93	0.9154	0.2240	0.0107
No log	93.96	282	0.5497	0.93	0.1607	0.7717	0.93	0.9154	0.2141	0.0107
No log	94.96	285	0.5487	0.93	0.1605	0.7711	0.93	0.9154	0.2138	0.0107
No log	95.96	288	0.5473	0.93	0.1601	0.7704	0.93	0.9154	0.2129	0.0106
No log	96.96	291	0.5467	0.93	0.1599	0.7701	0.93	0.9154	0.2124	0.0106
No log	97.96	294	0.5462	0.93	0.1597	0.7698	0.93	0.9154	0.2121	0.0106
No log	98.96	297	0.5458	0.93	0.1596	0.7694	0.93	0.9154	0.2120	0.0106
No log	99.96	300	0.5457	0.93	0.1595	0.7692	0.93	0.9154	0.2119	0.0106

Framework versions

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