gpt1B_DPO_model_ver2

This model is a fine-tuned version of AI-Sweden-Models/gpt-sw3-1.3b on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0069
• Rewards/chosen: 0.0482
• Rewards/rejected: -6.7989
• Rewards/accuracies: 1.0
• Rewards/margins: 6.8472
• Logps/rejected: -275.1611
• Logps/chosen: -112.8062
• Logits/rejected: -2.7230
• Logits/chosen: -2.9162

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-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.0089	0.79	200	0.0086	0.0829	-6.3343	1.0	6.4172	-270.5143	-112.4591	-2.7397	-2.9321
0.0031	1.59	400	0.0069	0.0482	-6.7989	1.0	6.8472	-275.1611	-112.8062	-2.7230	-2.9162

Framework versions

• PEFT 0.8.2
• Transformers 4.38.1
• Pytorch 2.2.0+cu118
• Datasets 2.17.1
• Tokenizers 0.15.2