qwen7b-dpo-s12base-dposeed-s42

<!-- 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. -->

qwen7b_dpo_from_sft_s12base_s42

This model is a fine-tuned version of 17Lab/qwen7b-full-sft-s12 on the assimilation_dpo_v1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0511
• Rewards/chosen: -0.2211
• Rewards/rejected: -5.3009
• Rewards/accuracies: 1.0
• Rewards/margins: 5.0798
• Logps/chosen: -13.0600
• Logps/rejected: -68.0014
• Logits/chosen: -1.9085
• Logits/rejected: -1.8952

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: 4
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• total_eval_batch_size: 4
• optimizer: Use OptimizerNames.ADAMW_TORCH 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: 0.1
• num_epochs: 1

Training results

Framework versions

• PEFT 0.18.1
• Transformers 5.6.0
• Pytorch 2.7.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2