Modelpublic

kanishka/smolm-autoreg-bpe-counterfactual-babylm-indef_articles_with_pl_nouns-removal-seed_1024-3e-4

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes29downloads

Model Card

smolm-autoreg-bpe-counterfactual-babylm-indefarticleswithplnouns-removal-seed_1024-3e-4

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

—Loss: 3.4219
—Accuracy: 0.4085

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: 0.0003
—trainbatchsize: 32
—evalbatchsize: 64
—seed: 1024
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 32000
—num_epochs: 20.0
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.7355	1.0	18601	3.9075	0.3465
3.4271	2.0	37202	3.6102	0.3750
3.2876	3.0	55803	3.5109	0.3901
3.2	4.0	74404	3.4616	0.3953
3.1388	5.0	93005	3.4213	0.4001
3.098	6.0	111606	3.3978	0.4025
3.056	7.0	130207	3.3816	0.4040
3.0187	8.0	148808	3.3687	0.4057
2.9967	9.0	167409	3.3774	0.4064
2.9669	10.0	186010	3.3665	0.4075
2.943	11.0	204611	3.3829	0.4083
2.9161	12.0	223212	3.3787	0.4082
2.8964	13.0	241813	3.3810	0.4084
2.8746	14.0	260414	3.3963	0.4083
2.8563	15.0	279015	3.3931	0.4083
2.8343	16.0	297616	3.4001	0.4087
2.8146	17.0	316217	3.4094	0.4083
2.7947	18.0	334818	3.4094	0.4086
2.7805	19.0	353419	3.4176	0.4084
2.7612	20.0	372020	3.4219	0.4085

Framework versions

—Transformers 4.35.0
—Pytorch 2.1.0+cu121
—Datasets 2.12.0
—Tokenizers 0.14.1