TheHassanSaud/gemma3_12B_dpo_full_checkpoints

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

gemma312Bdpofullcheckpoints

This model is a fine-tuned version of ramzanniaz331/gemma3-12b-2048-ds2-sft-v3 on the dpodataset1, the dpodataset2, the dpodataset3, the dpodataset4, the dpodataset5 and the dpodataset6 datasets. It achieves the following results on the evaluation set:

• —Loss: 0.6420
• —Rewards/chosen: 7.7457
• —Rewards/rejected: -2.4085
• —Rewards/accuracies: 0.8378
• —Rewards/margins: 10.1542
• —Logps/chosen: -3533.0
• —Logps/rejected: -4106.9712
• —Logits/chosen: 0.0052
• —Logits/rejected: 0.0054

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-06
• —trainbatchsize: 2
• —evalbatchsize: 1
• —seed: 42
• —distributed_type: multi-GPU
• —num_devices: 8
• —gradientaccumulationsteps: 16
• —totaltrainbatch_size: 256
• —totalevalbatch_size: 8
• —optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• —lrschedulertype: cosine
• —lrschedulerwarmup_steps: 0.05
• —num_epochs: 1.0

Training results

Framework versions

• —Transformers 5.0.0.dev0
• —Pytorch 2.9.1+cu128
• —Datasets 4.0.0
• —Tokenizers 0.22.2