categorization-finetuned-20220721-164940-pruned-20220803-123018

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: 0.5476
• Accuracy: 0.8558
• F1: 0.8539

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: 7e-06
• train_batch_size: 48
• eval_batch_size: 48
• seed: 314
• gradient_accumulation_steps: 6
• total_train_batch_size: 288
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.3421	0.51	2000	0.4324	0.8871	0.8864
0.3435	1.01	4000	0.4276	0.8885	0.8878
0.327	1.52	6000	0.4300	0.8891	0.8884
0.3299	2.02	8000	0.4266	0.8891	0.8885
0.3217	2.53	10000	0.4303	0.8881	0.8873
0.3347	3.04	12000	0.4291	0.8885	0.8879
0.3307	3.54	14000	0.4334	0.8873	0.8867
0.3537	4.05	16000	0.4340	0.8850	0.8844
0.3659	4.56	18000	0.4426	0.8828	0.8819
0.3933	5.06	20000	0.4485	0.8805	0.8796
0.4117	5.57	22000	0.4553	0.8779	0.8768
0.4501	6.07	24000	0.4734	0.8734	0.8725
0.4848	6.58	26000	0.4895	0.8690	0.8678
0.5182	7.09	28000	0.5137	0.8634	0.8617
0.54	7.59	30000	0.5165	0.8625	0.8610
0.5582	8.1	32000	0.5312	0.8591	0.8572
0.5728	8.61	34000	0.5382	0.8574	0.8556
0.5883	9.11	36000	0.5514	0.8553	0.8534
0.5942	9.62	38000	0.5563	0.8534	0.8512
0.6015	10.12	40000	0.5592	0.8536	0.8516
0.603	10.63	42000	0.5585	0.8533	0.8513
0.5972	11.14	44000	0.5585	0.8541	0.8520
0.5938	11.64	46000	0.5546	0.8548	0.8529
0.5882	12.15	48000	0.5515	0.8554	0.8535
0.5799	12.65	50000	0.5488	0.8561	0.8541
0.572	13.16	52000	0.5473	0.8566	0.8547
0.5718	13.67	54000	0.5468	0.8566	0.8547
0.5698	14.17	56000	0.5464	0.8566	0.8547
0.5696	14.68	58000	0.5464	0.8566	0.8547

Framework versions

• Transformers 4.18.0.dev0
• Pytorch 1.9.1+cu111
• Datasets 2.3.2
• Tokenizers 0.11.6