simpo

This model is a fine-tuned version of google/gemma-2-9b-it on the princeton-nlp/gemma2-ultrafeedback-armorm dataset. It achieves the following results on the evaluation set:

• Loss: 2.7459
• Rewards/chosen: -18.9664
• Rewards/rejected: -23.9949
• Rewards/accuracies: 0.7725
• Rewards/margins: 5.0285
• Logps/rejected: -2.3995
• Logps/chosen: -1.8966
• Logits/rejected: -14.4878
• Logits/chosen: -14.5537

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: 8e-07
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 16
• 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
2.7196	0.8594	400	2.7580	-18.9526	-23.9387	0.7705	4.9861	-2.3939	-1.8953	-14.4321	-14.5024

Framework versions

• Transformers 4.44.2
• Pytorch 2.7.0+cu128
• Datasets 2.18.0
• Tokenizers 0.19.1