mlrun/Meta-Llama-3-8B-fashion-v7.0

results

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: 0.6367
• Rewards/chosen: -0.0965
• Rewards/rejected: -0.4524
• Rewards/accuracies: 1.0
• Rewards/margins: 0.3560
• Logps/rejected: -2.2621
• Logps/chosen: -0.4823
• Logits/rejected: -1.8871
• Logits/chosen: -1.6815
• Nll Loss: 0.5673
• Log Odds Ratio: -0.0697
• Log Odds Chosen: 2.8182

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: 0.001
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen	Nll Loss	Log Odds Ratio	Log Odds Chosen
4.5921	0.2126	11	3.5043	-0.6803	-0.8039	1.0	0.1236	-4.0195	-3.4015	-3.2932	-3.2264	3.3934	-0.4430	0.6340
1.3234	0.4251	22	0.8683	-0.1360	-0.2818	1.0	0.1458	-1.4090	-0.6802	-2.6124	-2.3854	0.7738	-0.2790	1.1626
0.7333	0.6377	33	0.8013	-0.1269	-0.3679	1.0	0.2410	-1.8394	-0.6346	-2.3638	-2.1864	0.7403	-0.1581	1.8427
0.5916	0.8502	44	0.6367	-0.0965	-0.4524	1.0	0.3560	-2.2621	-0.4823	-1.8871	-1.6815	0.5673	-0.0697	2.8182

Framework versions

• PEFT 0.11.1
• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1