medllama3-v20-with-prefix_prompt

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8028
• Balanced Accuracy: 0.5960
• Accuracy: 0.5457

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Balanced Accuracy	Accuracy
1.7999	1.0	165	0.6813	0.6104	0.6187
1.7523	2.0	330	0.6623	0.5918	0.4087
2.1806	3.0	495	0.6799	0.6413	0.4269
1.3861	4.0	660	0.7232	0.6363	0.4452
1.6961	5.0	825	0.8122	0.6428	0.3721
1.4074	6.0	990	0.8685	0.6400	0.6735
1.3749	7.0	1155	0.6330	0.6409	0.6598
1.3455	8.0	1320	0.6396	0.6594	0.4886
1.3391	9.0	1485	0.6102	0.6464	0.5845
1.4026	10.0	1650	0.6464	0.6058	0.6142
1.2097	11.0	1815	0.7539	0.6151	0.6644
1.178	12.0	1980	0.6004	0.6539	0.6210
1.1876	13.0	2145	0.6344	0.6308	0.6279
1.1125	14.0	2310	0.6716	0.6598	0.6507
1.0532	15.0	2475	0.7006	0.6501	0.5548
1.0158	16.0	2640	0.6975	0.6445	0.6050
0.9504	17.0	2805	0.8028	0.5960	0.5457

Framework versions

• PEFT 0.14.0
• Transformers 4.48.1
• Pytorch 2.1.0
• Datasets 3.2.0
• Tokenizers 0.21.0