distilbert-classn-LinearAlg-finetuned-span-width-1

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: 0.7813
• Accuracy: 0.8016
• F1: 0.8033
• Precision: 0.8179
• Recall: 0.8016

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
5.1371	0.6849	50	2.4649	0.0714	0.0501	0.1218	0.0714
4.9171	1.3699	100	2.4518	0.0794	0.0533	0.1254	0.0794
4.8245	2.0548	150	2.3856	0.1032	0.0914	0.2028	0.1032
4.8193	2.7397	200	2.3362	0.1429	0.1561	0.2496	0.1429
4.6972	3.4247	250	2.3636	0.1032	0.1092	0.1813	0.1032
4.615	4.1096	300	2.3017	0.2302	0.2227	0.2808	0.2302
4.3801	4.7945	350	2.2082	0.2937	0.2870	0.3542	0.2937
4.209	5.4795	400	2.1003	0.3333	0.3097	0.4304	0.3333
3.916	6.1644	450	1.9879	0.4206	0.3987	0.4402	0.4206
3.5137	6.8493	500	1.7583	0.5397	0.5249	0.5419	0.5397
2.99	7.5342	550	1.5959	0.5476	0.5150	0.5721	0.5476
2.4576	8.2192	600	1.3742	0.6508	0.6430	0.7217	0.6508
2.0467	8.9041	650	1.2277	0.6825	0.6805	0.7112	0.6825
1.6407	9.5890	700	1.0865	0.6825	0.6764	0.7042	0.6825
1.1023	10.2740	750	0.9734	0.7302	0.7269	0.7686	0.7302
0.8708	10.9589	800	0.8830	0.7619	0.7565	0.7740	0.7619
0.7335	11.6438	850	0.8266	0.7698	0.7707	0.7922	0.7698
0.5333	12.3288	900	0.8078	0.7619	0.7603	0.7723	0.7619
0.389	13.0137	950	0.7685	0.7857	0.7874	0.8046	0.7857
0.3018	13.6986	1000	0.7756	0.7778	0.7829	0.8064	0.7778
0.2219	14.3836	1050	0.7737	0.7698	0.7667	0.7743	0.7698
0.1865	15.0685	1100	0.7674	0.7857	0.7846	0.7994	0.7857
0.1429	15.7534	1150	0.7750	0.7778	0.7796	0.7981	0.7778
0.1038	16.4384	1200	0.7642	0.7937	0.7964	0.8099	0.7937
0.0881	17.1233	1250	0.7472	0.8016	0.8051	0.8245	0.8016
0.0946	17.8082	1300	0.7663	0.7937	0.7974	0.8162	0.7937
0.0501	18.4932	1350	0.7531	0.7937	0.7928	0.8020	0.7937
0.0421	19.1781	1400	0.7649	0.7937	0.7951	0.8081	0.7937
0.051	19.8630	1450	0.7695	0.8016	0.8035	0.8164	0.8016
0.0297	20.5479	1500	0.7799	0.8016	0.8036	0.8212	0.8016
0.0229	21.2329	1550	0.7649	0.8016	0.8035	0.8160	0.8016
0.0232	21.9178	1600	0.7810	0.7937	0.7948	0.8117	0.7937
0.0309	22.6027	1650	0.7813	0.8016	0.8033	0.8179	0.8016
0.034	23.2877	1700	0.7838	0.8016	0.8038	0.8189	0.8016
0.0147	23.9726	1750	0.7822	0.8016	0.8033	0.8179	0.8016
0.0232	24.6575	1800	0.7813	0.8016	0.8033	0.8179	0.8016

Framework versions

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