smolm-mlm-bpe-unmask-seed_111

This model is a fine-tuned version of models/smolm-mlm/config.json on 5M words of American-English child-directed input. It achieves the following results on the evaluation set:

• Loss: 2.6956
• Accuracy: 0.4492

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
• train_batch_size: 64
• eval_batch_size: 512
• seed: 111
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 24000
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.5129	1.0	11938	3.4627	0.3523
3.319	2.0	23876	3.3322	0.3641
3.1577	3.0	35814	3.1841	0.3810
3.0357	4.0	47752	3.0588	0.3982
2.9606	5.0	59690	2.9535	0.4109
2.87	6.0	71628	2.8745	0.4221
2.7817	7.0	83566	2.8351	0.4284
2.7388	8.0	95504	2.7536	0.4417
2.6618	9.0	107442	2.7308	0.4424
2.6258	10.0	119380	2.6880	0.4522

Framework versions

• Transformers 4.32.1
• Pytorch 1.13.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3