search

Open3DA / LL3DA

[CVPR 2024] "LL3DA: Visual Interactive Instruction Tuning for Omni-3D Understanding, Reasoning, and Planning"; an interactive Large Language 3D Assistant.
https://ll3da.github.io/
MIT License
225 stars 9 forks source link

Result reproduction #20

Open Xusssssss opened 2 months ago

Xusssssss commented 2 months ago

Hello author, I am very interested in your work, but I can't find the corresponding implementation of the results in the paper in the code. May I ask which part of the code was used to obtain the experimental results in the following figure

2c55db5d20086fbfa3aaaaee93ae590
ch3cook-fdu commented 2 months ago

Please follow the following steps:

  1. Train the generalist model: bash scripts/opt-1.3b/train.generalist.sh or bash scripts-v0/opt-1.3b/train.generalist.sh.
  2. (optional) Fine-tune for ScanQA: bash scripts/opt-1.3b/tuning.scanqa.sh.
  3. Inference: bash scripts/opt-1.3b/eval.scanqa.sh.

The evaluations of the test set come from the ScanQA benchmark on EvalAI.