llama_3_1_8B_Pak_DPO

<!-- 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. -->

llama3_1_8B_dpo_newdata_full_checkpoints

This model is a fine-tuned version of ramzanniaz331/llama3-8b-full-sft-v3 on the dpo_dataset_1, the dpo_dataset_2, the dpo_dataset_3, the dpo_dataset_4, the dpo_dataset_5 and the dpo_dataset_6 datasets. It achieves the following results on the evaluation set:

• Loss: 0.1399
• Rewards/chosen: -4.0691
• Rewards/rejected: -62.5231
• Rewards/accuracies: 0.9375
• Rewards/margins: 58.4540
• Logps/chosen: -299.7252
• Logps/rejected: -1401.1368
• Logits/chosen: -1.4589
• Logits/rejected: -1.4782

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
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• total_eval_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_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