llama2-13b-dpo-lora-20231205-32

This model is a fine-tuned version of meta-llama/Llama-2-13b-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6891
• Rewards/chosen: 0.0224
• Rewards/rejected: 0.0145
• Rewards/accuracies: 0.5675
• Rewards/margins: 0.0079
• Logps/rejected: -245.1621
• Logps/chosen: -309.3195
• Logits/rejected: -1.2044
• Logits/chosen: -1.4437

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-07
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.691	1.0	121	0.6911	0.0078	0.0020	0.5040	0.0058	-245.2879	-309.4661	-1.2040	-1.4432
0.6895	2.0	242	0.6903	0.0202	0.0107	0.5317	0.0095	-245.2007	-309.3419	-1.2045	-1.4435
0.6889	3.0	363	0.6891	0.0224	0.0145	0.5675	0.0079	-245.1621	-309.3195	-1.2044	-1.4437

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu121
• Datasets 2.14.6
• Tokenizers 0.14.1