about the results of Interactive Evaluation

[yanan1116]

hi authors,

may I know have you include the experimental results and related code for the setting of VLM's Interactive Evaluation , under the chapter of 4.3. Comprehensive Ability of VLMs ?
I guess the Figure 5 radar chart is for Non-interactive Evaluation setting, right ?

thanks.

[Shiduo-zh]

Yes, you are right. The interactive evaluation of VLMs should use the motion planning functions to execute the plans VLM proposes, such as <Pick, 8>, which is time consuming. So we only report the non-inactive evaluation results for a reference.