scenario-kd-scr-ner-full-xlmr_data-univner_half44

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

• Loss: 238.5785
• Precision: 0.3821
• Recall: 0.2681
• F1: 0.3151
• Accuracy: 0.9380

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
446.8617	0.5828	500	369.9693	0.0	0.0	0.0	0.9241
345.828	1.1655	1000	339.4001	0.4474	0.0074	0.0145	0.9243
316.4135	1.7483	1500	320.2157	0.3593	0.0778	0.1279	0.9271
295.3783	2.3310	2000	303.7513	0.4250	0.0740	0.1261	0.9274
278.865	2.9138	2500	291.9803	0.3462	0.1311	0.1902	0.9310
265.0536	3.4965	3000	282.1734	0.3378	0.1561	0.2135	0.9319
252.7824	4.0793	3500	274.9576	0.3486	0.2065	0.2593	0.9342
243.1838	4.6620	4000	265.2098	0.3825	0.1775	0.2424	0.9352
235.3429	5.2448	4500	260.4372	0.3720	0.2352	0.2882	0.9369
227.8851	5.8275	5000	256.3334	0.3570	0.2534	0.2964	0.9365
221.9237	6.4103	5500	250.2192	0.3931	0.2375	0.2961	0.9383
217.836	6.9930	6000	245.7306	0.3999	0.2268	0.2894	0.9385
213.3779	7.5758	6500	242.3217	0.3961	0.2378	0.2972	0.9391
209.3609	8.1585	7000	241.0757	0.3846	0.2448	0.2992	0.9381
207.6172	8.7413	7500	239.1901	0.3905	0.2535	0.3074	0.9391
205.3707	9.3240	8000	239.5822	0.3728	0.2759	0.3171	0.9378
204.5786	9.9068	8500	238.5785	0.3821	0.2681	0.3151	0.9380

Framework versions

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