muril-base-cased-finetuned-code-mixed-DS

This model is a fine-tuned version of google/muril-base-cased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9319
• Accuracy: 0.6982
• Precision: 0.6327
• Recall: 0.6314
• F1: 0.6320

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: 32
• eval_batch_size: 32
• seed: 43
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.0542	1.98	248	0.9786	0.5976	0.3936	0.5454	0.4330
0.9307	3.97	496	0.8836	0.5996	0.4072	0.5604	0.4399
0.8323	5.95	744	0.8266	0.5996	0.5508	0.5720	0.4527
0.7554	7.94	992	0.8006	0.6318	0.5601	0.5838	0.5232
0.6821	9.92	1240	0.8777	0.6740	0.5929	0.5875	0.5836
0.6173	11.9	1488	0.8389	0.6640	0.5918	0.6031	0.5881
0.5552	13.89	1736	0.9003	0.6962	0.6240	0.6160	0.6191
0.4932	15.87	1984	0.8979	0.6982	0.6266	0.6231	0.6245
0.4446	17.86	2232	0.9104	0.7002	0.6310	0.6290	0.6298
0.4084	19.84	2480	0.9284	0.7002	0.6278	0.6255	0.6264
0.3763	21.82	2728	0.9228	0.7082	0.6436	0.6380	0.6398
0.3575	23.81	2976	0.9319	0.6982	0.6327	0.6314	0.6320

Framework versions

• Transformers 4.20.1
• Pytorch 1.10.1+cu111
• Datasets 2.3.2
• Tokenizers 0.12.1