ent_description

This model is a fine-tuned version of textattack/bert-base-uncased-CoLA on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6580
• Class Acc: 0.9292
• Class Rec: 0.8413
• Class Prec: 0.8281
• Class F1: 0.8346
• Class Mcc: 0.7896
• Feature Ham Loss: 0.2192
• Feature Rec: 0.8123
• Feature Prec: 0.8417
• Feature F1: 0.8242

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• 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
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Class Acc	Class Rec	Class Prec	Class F1	Class Mcc	Feature Ham Loss	Feature Rec	Feature Prec	Feature F1
No log	1.0	44	0.7838	0.8617	0.3571	0.9783	0.5233	0.5429	0.2437	0.8232	0.8093	0.8085
No log	2.0	88	0.6784	0.9258	0.7222	0.91	0.8053	0.7680	0.2626	0.7303	0.8361	0.7713
No log	3.0	132	0.7053	0.8870	0.9206	0.6705	0.7759	0.7186	0.2175	0.9294	0.7970	0.8501
No log	4.0	176	0.7280	0.8853	0.9286	0.6648	0.7748	0.7183	0.2226	0.9313	0.7921	0.8499
No log	5.0	220	0.6285	0.9376	0.7540	0.9406	0.8370	0.8065	0.2310	0.7869	0.8440	0.8071
No log	6.0	264	0.6029	0.9342	0.8095	0.8718	0.8395	0.7991	0.2293	0.8327	0.8206	0.8230
No log	7.0	308	0.6106	0.9410	0.8333	0.8824	0.8571	0.8205	0.2146	0.8810	0.8181	0.8371
No log	8.0	352	0.6012	0.9410	0.8333	0.8824	0.8571	0.8205	0.2104	0.8734	0.8242	0.8390
No log	9.0	396	0.5981	0.9376	0.8413	0.8618	0.8514	0.8120	0.2087	0.8632	0.8290	0.8387
No log	10.0	440	0.6213	0.9393	0.8175	0.8879	0.8512	0.8143	0.2323	0.7704	0.8491	0.8029
No log	11.0	484	0.6667	0.9258	0.8968	0.7847	0.8370	0.7922	0.2243	0.7856	0.8519	0.8156
0.6254	12.0	528	0.6419	0.9325	0.8730	0.8209	0.8462	0.8036	0.2201	0.8492	0.8222	0.8327
0.6254	13.0	572	0.6979	0.9258	0.9048	0.7808	0.8382	0.7940	0.2184	0.8772	0.8137	0.8426
0.6254	14.0	616	0.6541	0.9292	0.8492	0.8231	0.8359	0.7909	0.2209	0.7926	0.8504	0.8174
0.6254	15.0	660	0.6407	0.9342	0.8333	0.8537	0.8434	0.8019	0.2146	0.8340	0.8351	0.8298
0.6254	16.0	704	0.6448	0.9342	0.8016	0.8783	0.8382	0.7983	0.2171	0.8149	0.8424	0.8225
0.6254	17.0	748	0.6556	0.9309	0.7619	0.8972	0.8240	0.7854	0.2196	0.8181	0.8375	0.8218
0.6254	18.0	792	0.6642	0.9275	0.8571	0.8120	0.8340	0.7881	0.2163	0.8244	0.8389	0.8292
0.6254	19.0	836	0.6546	0.9376	0.8254	0.8739	0.8490	0.8102	0.2226	0.8066	0.8405	0.8203
0.6254	20.0	880	0.6580	0.9292	0.8413	0.8281	0.8346	0.7896	0.2192	0.8123	0.8417	0.8242

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0