sewd-classifier-aug

This model is a fine-tuned version of asapp/sew-d-tiny-100k-ft-ls100h on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3895
• Accuracy: 0.6280
• Precision: 0.6286
• Recall: 0.6280
• F1: 0.5911
• Binary: 0.7407

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.19	50	4.3686	0.0431	0.0158	0.0431	0.0125	0.2191
No log	0.38	100	4.1099	0.0512	0.0096	0.0512	0.0109	0.2987
No log	0.58	150	3.8483	0.0674	0.0273	0.0674	0.0229	0.3337
No log	0.77	200	3.6126	0.0809	0.0229	0.0809	0.0292	0.3544
No log	0.96	250	3.4229	0.1348	0.0686	0.1348	0.0691	0.3911
4.077	1.15	300	3.2957	0.1698	0.0840	0.1698	0.0859	0.4183
4.077	1.34	350	3.1928	0.2156	0.1085	0.2156	0.1245	0.4469
4.077	1.53	400	3.0884	0.2075	0.0961	0.2075	0.1141	0.4420
4.077	1.73	450	2.9780	0.2534	0.1994	0.2534	0.1676	0.4757
4.077	1.92	500	2.8808	0.2884	0.2057	0.2884	0.1981	0.4987
3.3556	2.11	550	2.7864	0.3100	0.2164	0.3100	0.2170	0.5156
3.3556	2.3	600	2.7081	0.3369	0.2348	0.3369	0.2450	0.5361
3.3556	2.49	650	2.6018	0.3423	0.2305	0.3423	0.2548	0.5391
3.3556	2.68	700	2.5388	0.3531	0.2630	0.3531	0.2644	0.5458
3.3556	2.88	750	2.4501	0.3558	0.2640	0.3558	0.2726	0.5493
2.9854	3.07	800	2.3623	0.4232	0.3298	0.4232	0.3373	0.5965
2.9854	3.26	850	2.2990	0.4232	0.3592	0.4232	0.3469	0.5951
2.9854	3.45	900	2.2174	0.4259	0.3381	0.4259	0.3490	0.5992
2.9854	3.64	950	2.1462	0.4555	0.3967	0.4555	0.3844	0.6199
2.9854	3.84	1000	2.0908	0.4447	0.3910	0.4447	0.3737	0.6102
2.6945	4.03	1050	2.0397	0.4528	0.3873	0.4528	0.3762	0.6191
2.6945	4.22	1100	1.9789	0.4906	0.4262	0.4906	0.4216	0.6445
2.6945	4.41	1150	1.9196	0.5229	0.4729	0.5229	0.4613	0.6671
2.6945	4.6	1200	1.8807	0.4960	0.4391	0.4960	0.4328	0.6493
2.6945	4.79	1250	1.8297	0.5175	0.4665	0.5175	0.4584	0.6633
2.6945	4.99	1300	1.8099	0.5175	0.4805	0.5175	0.4550	0.6633
2.4977	5.18	1350	1.7638	0.5283	0.4954	0.5283	0.4687	0.6709
2.4977	5.37	1400	1.7227	0.5283	0.4549	0.5283	0.4608	0.6701
2.4977	5.56	1450	1.6999	0.5472	0.5024	0.5472	0.4867	0.6833
2.4977	5.75	1500	1.6623	0.5445	0.5207	0.5445	0.4919	0.6822
2.4977	5.94	1550	1.6480	0.5499	0.5186	0.5499	0.4999	0.6860
2.3471	6.14	1600	1.6190	0.5714	0.5378	0.5714	0.5109	0.7011
2.3471	6.33	1650	1.6022	0.5687	0.5654	0.5687	0.5189	0.6992
2.3471	6.52	1700	1.5881	0.5660	0.5306	0.5660	0.5074	0.6973
2.3471	6.71	1750	1.5415	0.5795	0.5517	0.5795	0.5317	0.7067
2.3471	6.9	1800	1.5210	0.5849	0.5541	0.5849	0.5374	0.7105
2.2349	7.09	1850	1.4996	0.5984	0.5568	0.5984	0.5449	0.7199
2.2349	7.29	1900	1.4846	0.6065	0.6233	0.6065	0.5622	0.7256
2.2349	7.48	1950	1.4720	0.6065	0.6128	0.6065	0.5698	0.7256
2.2349	7.67	2000	1.4549	0.6011	0.6045	0.6011	0.5640	0.7218
2.2349	7.86	2050	1.4355	0.6307	0.6331	0.6307	0.5889	0.7426
2.1754	8.05	2100	1.4426	0.6119	0.6166	0.6119	0.5702	0.7294
2.1754	8.25	2150	1.4291	0.6226	0.6097	0.6226	0.5830	0.7369
2.1754	8.44	2200	1.4291	0.6119	0.6037	0.6119	0.5696	0.7294
2.1754	8.63	2250	1.4069	0.6307	0.6166	0.6307	0.5888	0.7426
2.1754	8.82	2300	1.4038	0.6199	0.6132	0.6199	0.5793	0.7350
2.138	9.01	2350	1.4045	0.6253	0.6265	0.6253	0.5848	0.7388
2.138	9.2	2400	1.4043	0.6226	0.6060	0.6226	0.5833	0.7369
2.138	9.4	2450	1.3902	0.6253	0.6109	0.6253	0.5846	0.7388
2.138	9.59	2500	1.3906	0.6253	0.6125	0.6253	0.5849	0.7388
2.138	9.78	2550	1.3915	0.6226	0.6075	0.6226	0.5838	0.7369
2.138	9.97	2600	1.3895	0.6280	0.6286	0.6280	0.5911	0.7407

Framework versions

• Transformers 4.38.2
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.15.1