categorization-finetuned-20220721-164940-distilled-20220811-074207

This model is a fine-tuned version of carted-nlp/categorization-finetuned-20220721-164940 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1499
• Accuracy: 0.8771
• F1: 0.8763

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 96
• seed: 314
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 1500
• num_epochs: 30.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.5644	0.56	2500	0.2739	0.7822	0.7774
0.2658	1.12	5000	0.2288	0.8159	0.8127
0.2307	1.69	7500	0.2082	0.8298	0.8273
0.2126	2.25	10000	0.1970	0.8389	0.8370
0.2012	2.81	12500	0.1888	0.8450	0.8433
0.1903	3.37	15000	0.1829	0.8496	0.8485
0.1846	3.94	17500	0.1783	0.8529	0.8511
0.1771	4.5	20000	0.1750	0.8548	0.8537
0.1726	5.06	22500	0.1727	0.8577	0.8564
0.1673	5.62	25000	0.1683	0.8602	0.8591
0.1648	6.19	27500	0.1675	0.8608	0.8597
0.1596	6.75	30000	0.1657	0.8630	0.8620
0.1563	7.31	32500	0.1635	0.8646	0.8639
0.154	7.87	35000	0.1613	0.8656	0.8647
0.1496	8.43	37500	0.1611	0.8666	0.8656
0.1496	9.0	40000	0.1598	0.8676	0.8669
0.1445	9.56	42500	0.1594	0.8681	0.8671
0.1435	10.12	45000	0.1588	0.8688	0.8679
0.1407	10.68	47500	0.1568	0.8703	0.8695
0.1382	11.25	50000	0.1564	0.8708	0.8700
0.1372	11.81	52500	0.1550	0.8720	0.8713
0.1344	12.37	55000	0.1559	0.8718	0.8708
0.1337	12.93	57500	0.1540	0.8735	0.8729
0.1303	13.5	60000	0.1541	0.8729	0.8721
0.1304	14.06	62500	0.1531	0.8735	0.8727
0.1274	14.62	65000	0.1535	0.8736	0.8727
0.1266	15.18	67500	0.1527	0.8750	0.8742
0.1251	15.74	70000	0.1525	0.8755	0.8748
0.1234	16.31	72500	0.1528	0.8753	0.8745
0.1229	16.87	75000	0.1516	0.8760	0.8753
0.121	17.43	77500	0.1523	0.8759	0.8752
0.1212	17.99	80000	0.1515	0.8760	0.8754
0.1185	18.56	82500	0.1514	0.8765	0.8757
0.1186	19.12	85000	0.1516	0.8766	0.8760
0.1172	19.68	87500	0.1506	0.8774	0.8767
0.1164	20.24	90000	0.1513	0.8770	0.8763
0.116	20.81	92500	0.1507	0.8774	0.8767
0.1145	21.37	95000	0.1507	0.8777	0.8770
0.1143	21.93	97500	0.1506	0.8776	0.8770
0.1131	22.49	100000	0.1507	0.8779	0.8772
0.1131	23.05	102500	0.1505	0.8779	0.8772
0.1123	23.62	105000	0.1506	0.8781	0.8774
0.1117	24.18	107500	0.1504	0.8783	0.8776
0.1118	24.74	110000	0.1503	0.8784	0.8777
0.1111	25.3	112500	0.1503	0.8783	0.8776
0.1111	25.87	115000	0.1502	0.8784	0.8777
0.1105	26.43	117500	0.1504	0.8783	0.8776
0.1105	26.99	120000	0.1502	0.8786	0.8779
0.1104	27.55	122500	0.1503	0.8786	0.8779
0.1096	28.12	125000	0.1502	0.8785	0.8779
0.1101	28.68	127500	0.1501	0.8786	0.8779
0.1101	29.24	130000	0.1502	0.8786	0.8779
0.1094	29.8	132500	0.1501	0.8786	0.8779

Framework versions

• Transformers 4.17.0
• Pytorch 1.11.0+cu113
• Datasets 2.3.2
• Tokenizers 0.11.6