Siyi Chen

Does sapien support generate mesh with texture of an actor?

19

Hi I'm trying to get mesh with texture after loading urdf object in SAPIEN. I find in docs that we could generate point clouds. Does SAPIEN support generating mesh with...

Why do we need negative sequences? I think contrastive learning do not need negative sequences, because negative pairs can define as two different frames in a video.

Hi, in the paper it is said that the trained model can predict 1 or 4 future frames conditioned on previous 4 frames. How is this achieved for predicting 1...

Code

1

Hi, would you mind sharing the code for self-supervised learning and fine-tuning? I find your work very interesting and would like to try out the implementation details. Thanks!

RGBDiff

5

Hi, when you extract RGBDiff, do you use frame at t+1 to minus frame at t without downsampling, and then do downsampling, or do you first do downsampling and then...

Finetune Question

1

Hi, when you fine tune the model, do you use a linear classifier, or an MLP classifier? Thank you!

Table 6 Question

2

Hi, in your paper table 6, why would having an MLP head instead of a linear head improve NN video retrieval accuracies? While doing NN video retrieval, do you train...

cumulative_logprob is None

1

Hi authors, When I try to run the vLLM demo, it turns out cumulative_logprob is None in the below line: `watermark_ppl = np.mean([-output.outputs[0].cumulative_logprob/len(output.outputs[0].token_ids) for output in outputs])` Do you have...

Hi, when I run `# Image Understanding CUDA_VISIBLE_DEVICES=0 python llavamini/eval/run_llava_mini.py \ --model-path ICTNLP/llava-mini-llama-3.1-8b \ --image-file llavamini/serve/examples/baby_cake.png \ --conv-mode llava_llama_3_1 --model-name "llava-mini" \ --query "What's the text on the cake?" `...