llama_8b_lima_8

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on the open_webui_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.7044

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: 8e-06
• train_batch_size: 3
• eval_batch_size: 3
• seed: 66
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 3
• total_train_batch_size: 18
• total_eval_batch_size: 6
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: polynomial
• lr_scheduler_warmup_steps: 25
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.6211	0.1	35	0.9107
0.6976	0.2	70	0.8526
0.5762	0.3	105	0.7999
0.5226	0.4	140	0.7717
0.5866	0.5	175	0.7433
0.5999	0.6	210	0.7398
0.6464	0.7	245	0.7204
0.552	0.8	280	0.7107
0.5657	0.9	315	0.7130
0.5687	1.0	350	0.7044

Framework versions

• Transformers 4.45.0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.20.0