zephyr-7b-dpo-full-ultrabin-high-bleu

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.6241
• Rewards/chosen: -0.3064
• Rewards/rejected: -0.5567
• Rewards/accuracies: 0.6953
• Rewards/margins: 0.2502
• Logps/rejected: -318.3273
• Logps/chosen: -293.2705
• Logits/rejected: -2.3523
• Logits/chosen: -2.3967

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: 8
• eval_batch_size: 8
• seed: 55
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• total_eval_batch_size: 64
• 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
0.6598	0.3484	50	0.6515	-0.1096	-0.2344	0.6602	0.1248	-286.0998	-273.5854	-2.5008	-2.5392
0.6378	0.6969	100	0.6302	-0.3350	-0.5719	0.6875	0.2369	-319.8471	-296.1287	-2.3948	-2.4383

Framework versions

• Transformers 4.44.0.dev0
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1