smolm-autoreg-bpe-babylm-no_aann-removal-1e-4

This model is a fine-tuned version of models/smolm-autoreg-bpe-babylm-no_aann-removal-1e-4/config.json on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.1735
• Accuracy: 0.4271

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: 64
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 32000
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
4.373	1.0	9181	4.4401	0.2946
3.7212	2.0	18362	3.7948	0.3561
3.4232	3.0	27543	3.5368	0.3834
3.2619	4.0	36724	3.3936	0.3974
3.1543	5.0	45905	3.3105	0.4061
3.0755	6.0	55086	3.2656	0.4119
3.0195	7.0	64267	3.2342	0.4155
2.9752	8.0	73448	3.2222	0.4174
2.9415	9.0	82629	3.1987	0.4200
2.9051	10.0	91810	3.1883	0.4221
2.8752	11.0	100991	3.1854	0.4227
2.8508	12.0	110172	3.1747	0.4241
2.8273	13.0	119353	3.1738	0.4243
2.8026	14.0	128534	3.1709	0.4251
2.7828	15.0	137715	3.1665	0.4259
2.7691	16.0	146896	3.1702	0.4261
2.7494	17.0	156077	3.1686	0.4265
2.7317	18.0	165258	3.1723	0.4269
2.7153	19.0	174439	3.1721	0.4269
2.6999	20.0	183620	3.1735	0.4271

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3