simpo-baseline-1e-7

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4908
• Rewards/chosen: -1.2930
• Rewards/rejected: -1.6583
• Rewards/accuracies: 0.6008
• Rewards/margins: 0.3653
• Logps/rejected: -0.6633
• Logps/chosen: -0.5172
• Logits/rejected: -1.2294
• Logits/chosen: -1.2648

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-06
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• total_eval_batch_size: 32
• 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
1.502	0.9981	467	1.4908	-1.2930	-1.6583	0.6008	0.3653	-0.6633	-0.5172	-1.2294	-1.2648

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.19.1