uf-tulu-2-7b-dpo-full

This model is a fine-tuned version of models/tulu-2-7b-uf-dpo on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

• Loss: 0.6557
• Rewards/chosen: -0.2228
• Rewards/rejected: -0.3409
• Rewards/accuracies: 0.7422
• Rewards/margins: 0.1181
• Logps/rejected: -359.4563
• Logps/chosen: -339.0808
• Logits/rejected: -1.0471
• Logits/chosen: -1.0577

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: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• 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
0.6765	0.4184	100	0.6735	-0.1002	-0.1550	0.7148	0.0548	-340.8675	-326.8218	-1.1563	-1.1550
0.6549	0.8368	200	0.6562	-0.2176	-0.3342	0.7305	0.1165	-358.7881	-338.5682	-1.0513	-1.0615

Framework versions

• Transformers 4.44.1
• Pytorch 2.1.2+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1