scenario-kd-scr-ner-half-xlmr_data-univner_full66

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: 121.6095
• Precision: 0.4270
• Recall: 0.3784
• F1: 0.4013
• Accuracy: 0.9476

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: 66
• 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
259.2613	0.2911	500	189.6813	0.0	0.0	0.0	0.9241
179.4566	0.5822	1000	173.2544	0.4545	0.0137	0.0266	0.9246
167.777	0.8732	1500	164.4018	0.3561	0.0361	0.0655	0.9256
160.7257	1.1643	2000	159.5333	0.0954	0.0042	0.0080	0.9243
155.3029	1.4554	2500	156.0974	0.2908	0.0283	0.0515	0.9254
152.3202	1.7465	3000	151.0953	0.2614	0.0521	0.0869	0.9266
148.4375	2.0375	3500	149.1951	0.4117	0.0326	0.0604	0.9256
144.8759	2.3286	4000	144.9364	0.2888	0.1091	0.1583	0.9284
141.9512	2.6197	4500	142.7550	0.3105	0.1097	0.1621	0.9288
139.8162	2.9108	5000	140.1311	0.3438	0.0981	0.1527	0.9289
136.3472	3.2019	5500	138.3547	0.2671	0.1955	0.2258	0.9327
134.5775	3.4929	6000	135.6351	0.2823	0.1747	0.2158	0.9340
132.1038	3.7840	6500	134.0777	0.2685	0.1773	0.2136	0.9349
130.6851	4.0751	7000	132.9280	0.2858	0.1840	0.2238	0.9359
128.5001	4.3662	7500	131.8978	0.3058	0.2001	0.2419	0.9369
127.3796	4.6573	8000	130.3655	0.3250	0.2151	0.2589	0.9378
126.5618	4.9483	8500	129.1083	0.3273	0.2332	0.2723	0.9383
125.2975	5.2394	9000	128.4492	0.3147	0.2560	0.2823	0.9396
123.5341	5.5305	9500	127.2300	0.3418	0.2580	0.2940	0.9405
122.698	5.8216	10000	126.8739	0.3390	0.2811	0.3073	0.9402
121.6237	6.1126	10500	125.7438	0.3739	0.3011	0.3336	0.9434
120.6456	6.4037	11000	125.2620	0.3606	0.3011	0.3282	0.9430
120.2335	6.6948	11500	124.5899	0.3759	0.3466	0.3606	0.9447
119.8109	6.9859	12000	123.9922	0.3920	0.3213	0.3532	0.9442
118.4398	7.2770	12500	123.5926	0.3971	0.3497	0.3719	0.9455
117.945	7.5680	13000	123.2072	0.4014	0.3308	0.3627	0.9453
117.9631	7.8591	13500	122.8442	0.4017	0.3556	0.3773	0.9458
117.3963	8.1502	14000	122.6162	0.3940	0.3769	0.3852	0.9464
116.5054	8.4413	14500	122.1343	0.4079	0.3718	0.3890	0.9471
116.5259	8.7324	15000	121.9603	0.4158	0.3528	0.3817	0.9470
116.4213	9.0234	15500	121.8525	0.4173	0.3718	0.3933	0.9470
115.7738	9.3145	16000	121.6751	0.4247	0.3767	0.3993	0.9476
115.8023	9.6056	16500	121.5823	0.4306	0.3855	0.4068	0.9479
115.8227	9.8967	17000	121.6095	0.4270	0.3784	0.4013	0.9476

Framework versions

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