Modelpublic

yigagilbert/t5_efficient_small_language_ID

sourceHugging Faceapache-2.0updated 10mo agoView on Hugging Face

0likes13downloads

Model Card

t5efficientsmalllanguageID

This model is a fine-tuned version of google/t5-efficient-small on the generator dataset. It achieves the following results on the evaluation set:

—Loss: 0.4285
—Accuracy: 0.6578
—F1 Macro: 0.5633
—F1 Weighted: 0.6050
—Precision Macro: 0.6452
—Recall Macro: 0.6124

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: 0.0005
—trainbatchsize: 64
—evalbatchsize: 64
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 128
—optimizer: Use adamwtorchfused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: cosinewithrestarts
—lrschedulerwarmup_steps: 1000
—training_steps: 60000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	F1 Weighted	Precision Macro	Recall Macro
0.3649	0.0083	500	0.8746	0.2458	0.1941	0.2013	0.2757	0.2370
0.1204	0.0167	1000	0.8914	0.3442	0.2543	0.2637	0.4155	0.3319
0.0788	0.025	1500	1.0181	0.3853	0.3001	0.3001	0.4832	0.3853
0.0771	0.0333	2000	0.5361	0.5775	0.4982	0.5166	0.5265	0.5569
0.0737	0.0417	2500	0.6765	0.5442	0.4678	0.4851	0.5409	0.5248
0.0399	0.05	3000	0.6103	0.5444	0.4692	0.4866	0.5858	0.5250
0.0557	0.0583	3500	0.4436	0.6128	0.5453	0.5655	0.6635	0.5909
0.0963	0.0667	4000	0.4755	0.6027	0.5328	0.5526	0.6001	0.5812
0.0282	0.075	4500	0.4607	0.6347	0.5728	0.5728	0.6121	0.6347
0.0386	0.0833	5000	0.5344	0.6501	0.5574	0.5781	0.6186	0.6269
0.0355	0.0917	5500	0.4191	0.6575	0.5793	0.6008	0.6199	0.6340
0.0244	0.1	6000	0.4040	0.6802	0.5880	0.6316	0.6406	0.6333
0.0331	0.1083	6500	0.4438	0.6517	0.6053	0.6053	0.7090	0.6517
0.0224	0.1167	7000	0.4869	0.6649	0.5878	0.6096	0.6689	0.6412
0.0263	0.125	7500	0.4285	0.6578	0.5633	0.6050	0.6452	0.6124

Framework versions

—Transformers 4.57.1
—Pytorch 2.9.0+cu128
—Datasets 4.3.0
—Tokenizers 0.22.1