distilbert-classn-LinearAlg-finetuned-span-width-5

This model is a fine-tuned version of dslim/distilbert-NER on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.4237
• Accuracy: 0.5476
• F1: 0.5578
• Precision: 0.5802
• Recall: 0.5476

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: 1e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
5.0508	0.6849	50	2.5065	0.0873	0.0655	0.0552	0.0873
5.0173	1.3699	100	2.4891	0.0873	0.0671	0.0598	0.0873
4.9755	2.0548	150	2.4689	0.0873	0.0594	0.0468	0.0873
4.9346	2.7397	200	2.4680	0.0873	0.0693	0.0656	0.0873
4.8569	3.4247	250	2.4446	0.0794	0.0691	0.0803	0.0794
4.7807	4.1096	300	2.4166	0.0873	0.0855	0.1281	0.0873
4.6991	4.7945	350	2.4031	0.0952	0.0940	0.1022	0.0952
4.4566	5.4795	400	2.3835	0.1508	0.1553	0.1795	0.1508
4.3468	6.1644	450	2.3874	0.1587	0.1524	0.1839	0.1587
4.2184	6.8493	500	2.3550	0.2063	0.2005	0.2522	0.2063
4.0597	7.5342	550	2.3082	0.2302	0.2093	0.2381	0.2302
3.8134	8.2192	600	2.2897	0.2302	0.2155	0.2873	0.2302
3.693	8.9041	650	2.2374	0.2540	0.2503	0.3564	0.2540
3.2232	9.5890	700	2.1660	0.2619	0.2491	0.3113	0.2619
3.0715	10.2740	750	2.0890	0.3175	0.3068	0.3467	0.3175
2.5457	10.9589	800	2.0022	0.3571	0.3422	0.3925	0.3571
2.2566	11.6438	850	1.9322	0.3413	0.3360	0.3502	0.3413
1.8691	12.3288	900	1.8635	0.3810	0.3572	0.3526	0.3810
1.6444	13.0137	950	1.7990	0.3889	0.3924	0.4215	0.3889
1.3832	13.6986	1000	1.7589	0.4524	0.4482	0.4583	0.4524
1.1667	14.3836	1050	1.7023	0.4365	0.4302	0.4431	0.4365
0.974	15.0685	1100	1.6077	0.4921	0.4881	0.4892	0.4921
0.8558	15.7534	1150	1.5825	0.4683	0.4645	0.4726	0.4683
0.7572	16.4384	1200	1.5705	0.4841	0.4796	0.4857	0.4841
0.6022	17.1233	1250	1.5357	0.5	0.4997	0.5102	0.5
0.5079	17.8082	1300	1.4927	0.5238	0.5295	0.5617	0.5238
0.4526	18.4932	1350	1.5055	0.4921	0.4949	0.5086	0.4921
0.4512	19.1781	1400	1.4643	0.5238	0.5270	0.5517	0.5238
0.3474	19.8630	1450	1.4326	0.5317	0.5402	0.5594	0.5317
0.2731	20.5479	1500	1.4435	0.5238	0.5310	0.5507	0.5238
0.2605	21.2329	1550	1.4289	0.5159	0.5249	0.5471	0.5159
0.235	21.9178	1600	1.4168	0.5556	0.5609	0.5747	0.5556
0.2219	22.6027	1650	1.4265	0.5317	0.5390	0.5591	0.5317
0.2373	23.2877	1700	1.4257	0.5476	0.5540	0.5734	0.5476
0.1764	23.9726	1750	1.4278	0.5397	0.5479	0.5698	0.5397
0.1967	24.6575	1800	1.4237	0.5476	0.5578	0.5802	0.5476

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.1
• Tokenizers 0.21.0