rubert-base-cased_neg

This model is a fine-tuned version of DeepPavlov/rubert-base-cased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6290
• Precision: 0.5977
• Recall: 0.6106
• F1: 0.6041
• Accuracy: 0.8995

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.11	50	0.6820	0.0	0.0	0.0	0.7747
No log	2.22	100	0.5532	0.0489	0.0212	0.0296	0.8040
No log	3.33	150	0.4231	0.1320	0.1004	0.1140	0.8397
No log	4.44	200	0.3736	0.2200	0.1873	0.2023	0.8576
No log	5.56	250	0.3511	0.3096	0.2606	0.2830	0.8713
No log	6.67	300	0.2976	0.3635	0.4498	0.4021	0.8833
No log	7.78	350	0.2713	0.3793	0.4942	0.4292	0.8925
No log	8.89	400	0.2569	0.4174	0.5753	0.4838	0.9016
No log	10.0	450	0.2599	0.4687	0.5772	0.5173	0.9031
0.389	11.11	500	0.2775	0.5008	0.5849	0.5396	0.9084
0.389	12.22	550	0.3123	0.4257	0.6641	0.5189	0.8834
0.389	13.33	600	0.3271	0.5036	0.5425	0.5223	0.9064
0.389	14.44	650	0.3390	0.5328	0.5174	0.5250	0.9016
0.389	15.56	700	0.3688	0.4788	0.6313	0.5445	0.8988
0.389	16.67	750	0.4260	0.4942	0.6583	0.5646	0.8997
0.389	17.78	800	0.3622	0.5217	0.6042	0.5599	0.9106
0.389	18.89	850	0.4054	0.5266	0.6120	0.5661	0.9105
0.389	20.0	900	0.3988	0.5070	0.6255	0.5601	0.9049
0.389	21.11	950	0.4422	0.5281	0.5444	0.5361	0.9093
0.0777	22.22	1000	0.4207	0.5635	0.5656	0.5645	0.9117
0.0777	23.33	1050	0.4721	0.5505	0.5792	0.5644	0.9130
0.0777	24.44	1100	0.4261	0.5379	0.6158	0.5743	0.9114
0.0777	25.56	1150	0.5339	0.6157	0.5290	0.5691	0.9
0.0777	26.67	1200	0.3761	0.4949	0.6544	0.5636	0.9036
0.0777	27.78	1250	0.4250	0.5650	0.5792	0.5720	0.9110
0.0777	28.89	1300	0.3790	0.5731	0.6429	0.6060	0.9193
0.0777	30.0	1350	0.5330	0.5942	0.5907	0.5924	0.9076
0.0777	31.11	1400	0.4419	0.5957	0.5888	0.5922	0.9180
0.0777	32.22	1450	0.5531	0.6008	0.5695	0.5847	0.9088
0.0362	33.33	1500	0.4544	0.5231	0.6564	0.5822	0.9135
0.0362	34.44	1550	0.4990	0.5695	0.5772	0.5733	0.9082
0.0362	35.56	1600	0.4040	0.5709	0.5753	0.5731	0.9086
0.0362	36.67	1650	0.3807	0.5989	0.6255	0.6119	0.9123
0.0362	37.78	1700	0.5088	0.5996	0.6100	0.6048	0.9139
0.0362	38.89	1750	0.4525	0.6151	0.5985	0.6067	0.9189
0.0362	40.0	1800	0.3787	0.6184	0.6100	0.6142	0.9211
0.0362	41.11	1850	0.3974	0.6097	0.5849	0.5970	0.9162
0.0362	42.22	1900	0.3944	0.5762	0.6274	0.6007	0.9148
0.0362	43.33	1950	0.3865	0.5124	0.6795	0.5842	0.9022
0.0264	44.44	2000	0.4583	0.5462	0.6274	0.5840	0.9169
0.0264	45.56	2050	0.4640	0.5635	0.6429	0.6005	0.9105
0.0264	46.67	2100	0.5028	0.5945	0.5830	0.5887	0.9128
0.0264	47.78	2150	0.3917	0.6267	0.6255	0.6261	0.9221
0.0264	48.89	2200	0.4833	0.6214	0.6274	0.6244	0.9138
0.0264	50.0	2250	0.4147	0.6130	0.6390	0.6257	0.9190
0.0264	51.11	2300	0.4455	0.6546	0.5927	0.6221	0.9185
0.0264	52.22	2350	0.4575	0.6138	0.6351	0.6243	0.9180
0.0264	53.33	2400	0.7707	0.3732	0.6815	0.4822	0.8354
0.0264	54.44	2450	0.4440	0.6015	0.6236	0.6123	0.9130
0.0248	55.56	2500	0.4815	0.5739	0.6448	0.6073	0.9124
0.0248	56.67	2550	0.3971	0.6204	0.6467	0.6333	0.9227
0.0248	57.78	2600	0.4770	0.6208	0.6100	0.6154	0.9193
0.0248	58.89	2650	0.5450	0.6699	0.5367	0.5959	0.9109
0.0248	60.0	2700	0.5033	0.5439	0.6332	0.5852	0.9019
0.0248	61.11	2750	0.5185	0.5187	0.6699	0.5847	0.9010
0.0248	62.22	2800	0.4277	0.6627	0.6371	0.6496	0.9194
0.0248	63.33	2850	0.4688	0.4869	0.6467	0.5556	0.9066
0.0248	64.44	2900	0.4779	0.6135	0.6313	0.6223	0.9153
0.0248	65.56	2950	0.5012	0.5852	0.6100	0.5974	0.9079
0.0232	66.67	3000	0.4788	0.5259	0.6081	0.5640	0.9052
0.0232	67.78	3050	0.4556	0.5726	0.6544	0.6108	0.9099
0.0232	68.89	3100	0.5026	0.608	0.5869	0.5972	0.9091
0.0232	70.0	3150	0.8153	0.3071	0.7143	0.4295	0.7567
0.0232	71.11	3200	0.4670	0.6169	0.6062	0.6115	0.9113
0.0232	72.22	3250	0.5249	0.5727	0.6313	0.6006	0.9068
0.0232	73.33	3300	0.4343	0.6085	0.6390	0.6234	0.9162
0.0232	74.44	3350	0.5067	0.6364	0.5811	0.6075	0.9135
0.0232	75.56	3400	0.4415	0.5812	0.6429	0.6104	0.9149
0.0232	76.67	3450	0.4052	0.5757	0.6313	0.6022	0.9137
0.0266	77.78	3500	0.5119	0.5233	0.5425	0.5327	0.9038
0.0266	78.89	3550	0.4689	0.5945	0.5888	0.5917	0.9145
0.0266	80.0	3600	0.3973	0.5609	0.6313	0.5940	0.9154
0.0266	81.11	3650	0.4848	0.5947	0.6486	0.6205	0.9181
0.0266	82.22	3700	0.4825	0.5877	0.6274	0.6069	0.9160
0.0266	83.33	3750	0.5193	0.5138	0.6467	0.5726	0.9
0.0266	84.44	3800	0.5344	0.5777	0.5811	0.5794	0.9107
0.0266	85.56	3850	0.5227	0.6591	0.5637	0.6077	0.9107
0.0266	86.67	3900	0.4490	0.5176	0.6255	0.5664	0.9097
0.0266	87.78	3950	0.6307	0.6464	0.5541	0.5967	0.9068
0.029	88.89	4000	0.4432	0.5667	0.5985	0.5822	0.9099
0.029	90.0	4050	0.4822	0.5148	0.6371	0.5695	0.9018
0.029	91.11	4100	0.4706	0.5966	0.6023	0.5994	0.9128

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.2