InternVL2.5-2B VL Reward Model LoRA

This repository contains the trained LoRA + scalar score head checkpoint for a VL reward model based on InternVL2.5-2B.

Result

  • Base model: InternVL2.5-2B
  • Training data: H4 RLAIF-V + RLHF-V mixed preference pairs
  • Evaluation benchmark: VLRewardBench
  • Score_base: 64.80%
  • Score_reward: 71.21%

Files

  • reward_lora_best.pt: best LoRA + score head checkpoint
  • internvl_reward.py: training/evaluation script used for this checkpoint
  • train_history.json: training history if available

VLRewardBench was used only for evaluation, not for training.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Hanzi7na/internvl25-2b-vlreward-lora

Adapter
(2)
this model