scenario-non-kd-scr-ner-half-xlmr_data-univner_full44

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3481
• Precision: 0.5331
• Recall: 0.5421
• F1: 0.5376
• Accuracy: 0.9563

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 44
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.3572	0.2910	500	0.3013	0.5	0.0752	0.1307	0.9280
0.2765	0.5821	1000	0.2546	0.3094	0.1636	0.2140	0.9317
0.2438	0.8731	1500	0.2356	0.3125	0.2232	0.2604	0.9349
0.218	1.1641	2000	0.2278	0.3203	0.2492	0.2803	0.9364
0.207	1.4552	2500	0.2202	0.3466	0.2864	0.3136	0.9383
0.1947	1.7462	3000	0.2112	0.3584	0.3033	0.3285	0.9397
0.184	2.0373	3500	0.2079	0.3458	0.3277	0.3365	0.9396
0.1621	2.3283	4000	0.2133	0.3726	0.3174	0.3428	0.9414
0.1652	2.6193	4500	0.2049	0.3700	0.3564	0.3631	0.9414
0.1597	2.9104	5000	0.2044	0.3646	0.3721	0.3683	0.9409
0.1483	3.2014	5500	0.2036	0.3927	0.3764	0.3844	0.9432
0.1426	3.4924	6000	0.2024	0.4004	0.3923	0.3963	0.9433
0.1364	3.7835	6500	0.2044	0.4041	0.3773	0.3902	0.9441
0.1327	4.0745	7000	0.2056	0.3913	0.3924	0.3918	0.9436
0.1219	4.3655	7500	0.2044	0.4298	0.3890	0.4084	0.9450
0.1209	4.6566	8000	0.1997	0.4286	0.4168	0.4226	0.9446
0.1187	4.9476	8500	0.2016	0.4285	0.4109	0.4195	0.9462
0.1061	5.2386	9000	0.2116	0.4328	0.4134	0.4229	0.9461
0.1035	5.5297	9500	0.2074	0.4547	0.4077	0.4299	0.9459
0.1029	5.8207	10000	0.2075	0.4410	0.4229	0.4317	0.9472
0.0977	6.1118	10500	0.2040	0.4326	0.4587	0.4452	0.9469
0.0867	6.4028	11000	0.2035	0.4417	0.4702	0.4555	0.9474
0.0848	6.6938	11500	0.2004	0.4452	0.4639	0.4544	0.9483
0.0804	6.9849	12000	0.2074	0.4549	0.4435	0.4492	0.9497
0.0719	7.2759	12500	0.2106	0.4675	0.4601	0.4638	0.9503
0.0696	7.5669	13000	0.2130	0.4635	0.4604	0.4619	0.9509
0.068	7.8580	13500	0.2106	0.4703	0.4793	0.4748	0.9514
0.0603	8.1490	14000	0.2190	0.4918	0.4738	0.4827	0.9521
0.056	8.4400	14500	0.2139	0.4830	0.4991	0.4909	0.9516
0.0589	8.7311	15000	0.2183	0.5031	0.4803	0.4914	0.9527
0.054	9.0221	15500	0.2242	0.4808	0.4923	0.4865	0.9527
0.047	9.3132	16000	0.2215	0.4873	0.5001	0.4936	0.9527
0.048	9.6042	16500	0.2237	0.4772	0.4999	0.4883	0.9527
0.0463	9.8952	17000	0.2283	0.4896	0.4930	0.4913	0.9526
0.0406	10.1863	17500	0.2338	0.5064	0.5019	0.5042	0.9539
0.0376	10.4773	18000	0.2348	0.5104	0.5009	0.5056	0.9540
0.0406	10.7683	18500	0.2395	0.5153	0.4975	0.5062	0.9538
0.0349	11.0594	19000	0.2456	0.5159	0.5070	0.5114	0.9542
0.0322	11.3504	19500	0.2482	0.5097	0.5043	0.5070	0.9539
0.0319	11.6414	20000	0.2518	0.5006	0.5144	0.5074	0.9546
0.0325	11.9325	20500	0.2532	0.4941	0.4983	0.4962	0.9531
0.0276	12.2235	21000	0.2593	0.5146	0.5095	0.5120	0.9543
0.0275	12.5146	21500	0.2537	0.5151	0.5161	0.5156	0.9545
0.0261	12.8056	22000	0.2537	0.5109	0.5327	0.5215	0.9547
0.0262	13.0966	22500	0.2634	0.5205	0.5149	0.5177	0.9547
0.0232	13.3877	23000	0.2666	0.5059	0.5197	0.5127	0.9548
0.0217	13.6787	23500	0.2718	0.5140	0.5178	0.5159	0.9540
0.0227	13.9697	24000	0.2668	0.5105	0.5272	0.5187	0.9550
0.0193	14.2608	24500	0.2743	0.5189	0.5309	0.5249	0.9552
0.019	14.5518	25000	0.2737	0.5063	0.5410	0.5231	0.9545
0.0207	14.8428	25500	0.2740	0.5243	0.5347	0.5295	0.9555
0.0188	15.1339	26000	0.2783	0.5191	0.5325	0.5257	0.9553
0.0171	15.4249	26500	0.2837	0.5338	0.5064	0.5197	0.9555
0.0173	15.7159	27000	0.2878	0.5355	0.5106	0.5227	0.9559
0.0168	16.0070	27500	0.2853	0.5267	0.5180	0.5223	0.9555
0.0144	16.2980	28000	0.2925	0.5015	0.5523	0.5257	0.9547
0.0153	16.5891	28500	0.2924	0.5294	0.5232	0.5263	0.9558
0.015	16.8801	29000	0.2882	0.5409	0.5337	0.5373	0.9562
0.0142	17.1711	29500	0.3034	0.5063	0.5256	0.5157	0.9551
0.0127	17.4622	30000	0.3021	0.5286	0.5284	0.5285	0.9557
0.0131	17.7532	30500	0.2979	0.5189	0.5422	0.5303	0.9553
0.0119	18.0442	31000	0.3031	0.5292	0.5262	0.5277	0.9558
0.011	18.3353	31500	0.3102	0.5094	0.5307	0.5198	0.9554
0.012	18.6263	32000	0.3057	0.5285	0.5240	0.5263	0.9554
0.0114	18.9173	32500	0.3095	0.5347	0.5296	0.5322	0.9560
0.0105	19.2084	33000	0.3123	0.5281	0.5288	0.5284	0.9558
0.0108	19.4994	33500	0.3125	0.5267	0.5418	0.5341	0.9560
0.0097	19.7905	34000	0.3116	0.5363	0.5208	0.5285	0.9559
0.0087	20.0815	34500	0.3124	0.5250	0.5367	0.5308	0.9556
0.0093	20.3725	35000	0.3195	0.5349	0.5252	0.5300	0.9560
0.0088	20.6636	35500	0.3207	0.5356	0.5296	0.5326	0.9560
0.009	20.9546	36000	0.3168	0.5333	0.5447	0.5389	0.9560
0.0078	21.2456	36500	0.3190	0.5273	0.5418	0.5344	0.9557
0.0076	21.5367	37000	0.3274	0.5436	0.5372	0.5403	0.9566
0.0077	21.8277	37500	0.3234	0.5490	0.5338	0.5413	0.9565
0.0086	22.1187	38000	0.3280	0.5258	0.5317	0.5287	0.9558
0.0064	22.4098	38500	0.3255	0.5273	0.5392	0.5332	0.9558
0.0074	22.7008	39000	0.3256	0.5302	0.5438	0.5369	0.9562
0.007	22.9919	39500	0.3271	0.5301	0.5483	0.5390	0.9561
0.0069	23.2829	40000	0.3330	0.5315	0.5413	0.5364	0.9561
0.0057	23.5739	40500	0.3372	0.5316	0.5429	0.5372	0.9566
0.0065	23.8650	41000	0.3376	0.5311	0.5314	0.5312	0.9561
0.006	24.1560	41500	0.3363	0.5352	0.5455	0.5403	0.9564
0.0057	24.4470	42000	0.3365	0.5383	0.5390	0.5386	0.9562
0.0059	24.7381	42500	0.3391	0.5354	0.5383	0.5369	0.9562
0.0059	25.0291	43000	0.3441	0.5348	0.5392	0.5370	0.9564
0.0053	25.3201	43500	0.3444	0.5282	0.5361	0.5321	0.9559
0.0055	25.6112	44000	0.3453	0.5334	0.5338	0.5336	0.9563
0.0052	25.9022	44500	0.3434	0.5408	0.5379	0.5393	0.9564
0.0048	26.1932	45000	0.3431	0.5376	0.5390	0.5383	0.9562
0.0047	26.4843	45500	0.3475	0.5350	0.5396	0.5373	0.9560
0.0049	26.7753	46000	0.3487	0.5296	0.5400	0.5348	0.9562
0.0049	27.0664	46500	0.3468	0.5305	0.5454	0.5378	0.9562
0.0043	27.3574	47000	0.3479	0.5297	0.5462	0.5378	0.9561
0.0048	27.6484	47500	0.3481	0.5331	0.5421	0.5376	0.9563

Framework versions

• Transformers 4.44.2
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.19.1