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

This model is a fine-tuned version of haryoaw/scenario-TCR-NER_data-univner_full on the None dataset. It achieves the following results on the evaluation set:

• Loss: 120.6807
• Precision: 0.4293
• Recall: 0.4089
• F1: 0.4189
• Accuracy: 0.9498

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: 8
• eval_batch_size: 32
• seed: 44
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
257.123	0.2911	500	190.7746	0.0	0.0	0.0	0.9241
179.8008	0.5822	1000	172.0475	0.4174	0.0215	0.0409	0.9250
166.9722	0.8732	1500	168.3299	0.1982	0.0700	0.1034	0.9262
160.8387	1.1643	2000	159.9843	0.3943	0.0140	0.0270	0.9247
156.1421	1.4554	2500	155.4852	0.2372	0.0975	0.1382	0.9274
151.8715	1.7465	3000	150.9896	0.2908	0.0581	0.0969	0.9266
147.6463	2.0375	3500	147.4566	0.2915	0.0597	0.0991	0.9266
143.651	2.3286	4000	144.4443	0.2725	0.0756	0.1184	0.9278
141.8507	2.6197	4500	141.7467	0.2457	0.1495	0.1859	0.9296
138.4918	2.9108	5000	139.3207	0.2607	0.1085	0.1532	0.9298
135.9046	3.2019	5500	137.1947	0.2604	0.1374	0.1798	0.9316
133.8725	3.4929	6000	135.3692	0.2597	0.1658	0.2024	0.9329
131.6288	3.7840	6500	134.0009	0.2604	0.1837	0.2154	0.9352
130.0139	4.0751	7000	132.1269	0.2941	0.2082	0.2438	0.9370
128.3934	4.3662	7500	131.1457	0.3030	0.2125	0.2498	0.9366
127.0388	4.6573	8000	129.8376	0.3060	0.2624	0.2826	0.9389
125.8219	4.9483	8500	128.8504	0.3098	0.2495	0.2764	0.9388
124.1103	5.2394	9000	127.8314	0.3401	0.2832	0.3091	0.9410
122.8369	5.5305	9500	127.0276	0.3409	0.3168	0.3284	0.9415
122.6045	5.8216	10000	126.1700	0.3644	0.2913	0.3238	0.9431
121.3109	6.1126	10500	125.1837	0.3703	0.3202	0.3434	0.9442
120.3202	6.4037	11000	124.8068	0.3985	0.3040	0.3449	0.9443
119.4787	6.6948	11500	124.0347	0.3937	0.3298	0.3589	0.9454
118.5185	6.9859	12000	123.6216	0.4032	0.3298	0.3629	0.9459
117.9392	7.2770	12500	123.3605	0.4185	0.3352	0.3722	0.9466
117.1901	7.5680	13000	122.6046	0.4163	0.3518	0.3813	0.9477
116.9696	7.8591	13500	122.1267	0.4206	0.3735	0.3957	0.9481
116.1554	8.1502	14000	121.7281	0.4289	0.3649	0.3943	0.9486
115.7022	8.4413	14500	121.4247	0.4332	0.3963	0.4140	0.9491
115.7436	8.7324	15000	121.2025	0.4291	0.4087	0.4187	0.9496
115.2168	9.0234	15500	121.0645	0.4274	0.3981	0.4122	0.9495
114.7331	9.3145	16000	120.8095	0.4204	0.3956	0.4076	0.9494
114.5054	9.6056	16500	120.7807	0.4369	0.3929	0.4137	0.9500
114.9341	9.8967	17000	120.6807	0.4293	0.4089	0.4189	0.9498

Framework versions

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