Thank you for your outstanding work. I've noticed that the recently released Qwen3VL demonstrates significantly enhanced base model capabilities compared to Qwen2.5VL. Have you considered training a more powerful evaluation model using Qwen3VL as the base model and open-sourcing it?
Thank you for your outstanding work. I've noticed that the recently released Qwen3VL demonstrates significantly enhanced base model capabilities compared to Qwen2.5VL. Have you considered training a more powerful evaluation model using Qwen3VL as the base model and open-sourcing it?