zephyr-7b-dpo-qlora

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-qlora on the updated and the original datasets. It achieves the following results on the evaluation set:

• Loss: 0.5735
• Rewards/chosen: -0.6770
• Rewards/rejected: -1.1070
• Rewards/accuracies: 0.6940
• Rewards/margins: 0.4300
• Logps/rejected: -351.8942
• Logps/chosen: -331.1508
• Logits/rejected: -1.4599
• Logits/chosen: -1.7015

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-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• total_eval_batch_size: 8
• 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
0.6269	0.32	100	0.6269	-0.2377	-0.4431	0.6820	0.2054	-285.4985	-287.2169	-2.2566	-2.3666
0.6332	0.64	200	0.5821	-0.5909	-0.9588	0.7060	0.3679	-337.0687	-322.5442	-1.6871	-1.8938
0.5648	0.96	300	0.5735	-0.6770	-1.1070	0.6940	0.4300	-351.8942	-331.1508	-1.4599	-1.7015

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.2.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2