ninagroot/Llama-360M-finaltest

Llama-360M

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.8245

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
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 300
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
8.6417	1.0	3	8.5751
8.3908	2.0	6	8.3473
7.9583	3.0	9	7.9814
7.3598	4.0	12	7.5011
6.7468	5.0	15	6.9942
6.3345	6.0	18	6.6309
6.0489	7.0	21	6.3987
5.9651	8.0	24	6.2101
5.7683	9.0	27	5.9691
5.3051	10.0	30	5.5791
4.6791	11.0	33	5.1445
4.3962	12.0	36	4.8859
4.0007	13.0	39	4.7013
3.9473	14.0	42	4.4994
3.5486	15.0	45	4.3178
3.3243	16.0	48	4.1587
3.1305	17.0	51	4.0505
2.8703	18.0	54	3.9467
2.7661	19.0	57	3.8780
2.7976	20.0	60	3.8245

Framework versions

• Transformers 4.39.1
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0