smolm-autoreg-bpe-babylm-no_aann-removal-3e-3

This model is a fine-tuned version of models/smolm-autoreg-bpe-babylm-no_aann-removal-3e-3/config.json on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.1435
• Accuracy: 0.4338

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: 0.003
• 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
3.4382	1.0	9181	3.5341	0.3795
3.2235	2.0	18362	3.3513	0.3988
3.1337	3.0	27543	3.2862	0.4073
3.065	4.0	36724	3.2328	0.4130
2.9911	5.0	45905	3.1884	0.4188
2.9331	6.0	55086	3.1616	0.4229
2.8872	7.0	64267	3.1435	0.4256
2.8504	8.0	73448	3.1346	0.4271
2.8204	9.0	82629	3.1270	0.4283
2.7873	10.0	91810	3.1158	0.4303
2.7587	11.0	100991	3.1263	0.4309
2.7354	12.0	110172	3.1099	0.4322
2.7105	13.0	119353	3.1115	0.4329
2.6833	14.0	128534	3.1166	0.4329
2.6605	15.0	137715	3.1237	0.4330
2.6432	16.0	146896	3.1246	0.4334
2.6192	17.0	156077	3.1203	0.4341
2.5964	18.0	165258	3.1337	0.4338
2.5737	19.0	174439	3.1365	0.4339
2.5522	20.0	183620	3.1435	0.4338

Framework versions

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