Modelpublic

andreas-in-den-bergen/distilbert-base-uncased-finetuned-emotion

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face

0likes6downloads

Model Card

distilbert-base-uncased-finetuned-emotion

This model is a fine-tuned version of distilbert-base-uncased on the emotion dataset. It achieves the following results on the evaluation set:

—Loss: 0.5889
—Accuracy: 0.86
—F1: 0.8571

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
—trainbatchsize: 64
—evalbatchsize: 64
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.2434	1.0	25	1.0831	0.605	0.4881
0.9611	2.0	50	0.9177	0.705	0.6476
0.7196	3.0	75	0.7741	0.735	0.7132
0.4981	4.0	100	0.6540	0.785	0.7715
0.3496	5.0	125	0.5721	0.83	0.8245
0.2435	6.0	150	0.5500	0.83	0.8266
0.1727	7.0	175	0.4813	0.86	0.8580
0.1253	8.0	200	0.5203	0.85	0.8479
0.0909	9.0	225	0.5427	0.83	0.8255
0.066	10.0	250	0.5391	0.845	0.8425
0.0474	11.0	275	0.5305	0.86	0.8575
0.0382	12.0	300	0.5264	0.855	0.8520
0.0306	13.0	325	0.5419	0.87	0.8673
0.0292	14.0	350	0.5868	0.855	0.8512
0.0224	15.0	375	0.5751	0.85	0.8464
0.0192	16.0	400	0.5837	0.85	0.8464
0.016	17.0	425	0.5852	0.855	0.8523
0.0151	18.0	450	0.5880	0.855	0.8523
0.0146	19.0	475	0.5912	0.86	0.8571
0.0131	20.0	500	0.5889	0.86	0.8571

Framework versions

—Transformers 4.35.2
—Pytorch 2.1.0+cu118
—Datasets 2.15.0
—Tokenizers 0.15.0