hf_llama3_lora

This model is a fine-tuned version of microsoft/Phi-3-mini-4k-instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2972

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 5
• gradient_accumulation_steps: 32
• total_train_batch_size: 640
• total_eval_batch_size: 20
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
1.4042	0.1862	500	1.3990
1.3445	0.3723	1000	1.3608
1.291	0.5585	1500	1.3493
1.264	0.7446	2000	1.3381
1.2438	0.9308	2500	1.3257
1.2333	1.1169	3000	1.3242
1.2084	1.3031	3500	1.3167
1.2227	1.4892	4000	1.3178
1.2151	1.6754	4500	1.3092
1.2114	1.8615	5000	1.3060
1.1645	2.0477	5500	1.3068
1.1793	2.2338	6000	1.3026
1.1809	2.4200	6500	1.3014
1.1934	2.6061	7000	1.2935
1.175	2.7923	7500	1.2953
1.1629	2.9784	8000	1.2954
1.1559	3.1646	8500	1.2972

Framework versions

• PEFT 0.9.0
• Transformers 4.43.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.18.0
• Tokenizers 0.19.1