llama-3-8b-instruct-gapo-v2-rouge2-beta10-1minus-gamma0.3-rerun

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the princeton-nlp/llama3-ultrafeedback-armorm dataset. It achieves the following results on the evaluation set:

• Loss: 1.2597
• Rewards/chosen: -17.5481
• Rewards/rejected: -23.3529
• Rewards/accuracies: 0.8415
• Rewards/margins: 5.8049
• Logps/rejected: -2.3353
• Logps/chosen: -1.7548
• Logits/rejected: -1.4709
• Logits/chosen: -1.4625

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-06
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• 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
1.2554	0.8550	400	1.2597	-17.5481	-23.3529	0.8415	5.8049	-2.3353	-1.7548	-1.4709	-1.4625

Framework versions

• Transformers 4.44.2
• Pytorch 2.2.0
• Datasets 2.21.0
• Tokenizers 0.19.1