hello, thanks for your great work. i work on on-board VLM, code ability is important for agentic mllm, for text search/image search is not key feature for on-board VLM, do you test just inject the code tool ability for VLM?and why need search tool, can code ability instead of search ability for agentic VLM?
hello, thanks for your great work. i work on on-board VLM, code ability is important for agentic mllm, for text search/image search is not key feature for on-board VLM, do you test just inject the code tool ability for VLM?and why need search tool, can code ability instead of search ability for agentic VLM?