FEAT: support InternVL series models, e.g. InternVL-Chat-V1.5
Is your feature request related to a problem? Please describe
Yes, the current platform lacks support for the cutting-edge InternVL-Chat-V1.5 model, which limits the variety of state-of-the-art models available for users to explore and leverage in their projects. Despite its outstanding performance on various benchmarks, it is not integrated into the existing framework, making it inaccessible to the community for both research and practical applications.
Describe the solution you'd like
I would like to request the addition of support for the InternVL-Chat-V1.5 model in the platform, enabling users to easily access, fine-tune, and utilize this powerful language model. This would include:
- Providing the model's weights and configuration files in the appropriate format for the platform, ensuring seamless integration.
- Updating the model hub or library to include InternVL-Chat-V1.5, allowing users to easily search and load the model.
By incorporating InternVL-Chat-V1.5, the platform would become even more comprehensive, offering researchers and developers access to the best performing models in various natural language processing tasks.
Describe alternatives you've considered
One alternative solution would be for users to manually adapt the model to work with the platform, which might involve significant efforts in code modification, file conversion, and compatibility checks. This, however, would defeat the purpose of having a unified platform that streamlines the process of working with various models.
Another alternative would be to rely on third-party repositories or custom implementations, which may lack the stability, consistency, and maintenance offered by the official platform. This could lead to inconsistencies and difficulties in reproducing results or troubleshooting issues.
Additional context
The OpenVLM Leaderboard ranks InternVL-Chat-V1.5 as the top model, indicating its superiority in various tasks. By integrating this model into the platform, users will have access to cutting-edge technology, enabling them to develop more advanced and accurate applications. This will ultimately improve the platform's overall functionality and position it at the forefront of AI innovation.