llama3-8B-EIP-8bit-lora

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

• Loss: 1.5218

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.4743	0.0433	100	1.7806
1.6838	0.0866	200	1.6540
1.6055	0.1299	300	1.6010
1.5809	0.1732	400	1.5646
1.5472	0.2165	500	1.5350
1.5206	0.2599	600	1.5218
1.5358	0.3032	700	1.5218
1.5102	0.3465	800	1.5218
1.552	0.3898	900	1.5218
1.5354	0.4331	1000	1.5218
1.5269	0.4764	1100	1.5218
1.5202	0.5197	1200	1.5218
1.5434	0.5630	1300	1.5218
1.5325	0.6063	1400	1.5218
1.5307	0.6496	1500	1.5218
1.5287	0.6929	1600	1.5218
1.5277	0.7362	1700	1.5218
1.5176	0.7796	1800	1.5218
1.5268	0.8229	1900	1.5218
1.5306	0.8662	2000	1.5218
1.5309	0.9095	2100	1.5218
1.5484	0.9528	2200	1.5218
1.51	0.9961	2300	1.5218

Framework versions

• PEFT 0.11.1
• Transformers 4.42.3
• Pytorch 2.1.0a0+4136153
• Datasets 2.20.0
• Tokenizers 0.19.1