Yichi Zhang

Running multiple instances on a single GPU occationally get empty frame

1

Hi, I am trying to run multiple AI2THOR instances on a single NVIDIA A6000 GPU. I have correctly configured the x server, and successfully tested my code under the single...

Run AI2THOR in clusters as a non-root user

4

Hi, I want to use AI2THOR on my university's cluster, where I am not allowed to start an x server as a non-root user. The docker image is also forbiddened...

Change the color or material for a specific object

1

Hi, Is there a way to select or randomize the color/material for a specific object instead of the whole scene? For example, is it possible to colorize a white plate...

zero-shot multi-selection QA

1

Hi, Is there a quick way to perform multi-selection QA over a provided answer candidate set using the pretrained checkpoints? Thanks!

The same action prediction gets different evaluation metrics in different runs

3

Hi, I ran the baseline ET model and found that two different runs get significantly different evaluation metrics. (might relate to this issue #10) Run1: ``` SR: 77/608 = 0.127...

Implementations of the rule-based TATC baseline

1

Hi, Thanks for releasing such a high-quality and large-scale dataset. For the Two-Agent Task Completion (TATC) benchmark, the paper proposes a rule-based approach for both agents as the baseline. Do...

Hi, Is there a way to get the high-fidelity game video streaming from the web UI as the agent's observation? Currently the images from the `ArenaOrchestrator` are discrete and low-resolution...

Video FPS

2

Hi, Thanks for the great work. May I ask what is the FPS used in extracting the frames? Thanks Best