[Mobile NeRF] Question: Training on a single GPU about jax3d HOT 10 OPEN

I also try with lego model in single 3090 successfully

from jax3d.

We had a 8x GPU machine and always used that. Yes, it's possible that for simpler scenes you can get by with not as many accelerators.

Hi, is it possible to combine instant-ngp with mobile-nerf?

from jax3d.

I also try with lego model in single 3090 successfully

Hi ,
would like to know how long it will take to finish the whole training process ? other scenes would also train successfully ? I used the single RTX3080 and it failed because of OOM

from jax3d.

I also try with lego model in single 3090 successfully

Hi , would like to know how long it will take to finish the whole training process ? other scenes would also train successfully ? I used the single RTX3080 and it failed because of OOM

Almost same time(maybe a bit less) described in the paper,
that's the most strange part, cause I don't got 8 A100....

from jax3d.

I try in WSL btw

from jax3d.

We had a 8x GPU machine and always used that.
Yes, it's possible that for simpler scenes you can get by with not as many accelerators.

from jax3d.

Yes. It is just a bit harder to implement than the current MLP version.

from jax3d.

我也尝试在单3090成功使用的型号lego

嗨，想知道完成整个培训过程需要多长时间？其他场景也会成功训练？我使用了单个RTX3080，但由于OOM而失败

I used 6 v100 GPUs，but had the same problem：

from jax3d.

set test_batch_size to smaller，Now it can run。

from jax3d.

Hello! First of all, congratulations on such an amazing paper and thank you very much for making the code public. I had a question regarding the requirements for training, is there a reason why one shouldn't attempt to train with a single GPU? It says that 8 V100 GPUs are required to "successfully train the model" However, I tested it with a single RTX3090 GPU and the results seem to be around the same as the paper (i.e. Same mesh quality as in the paper and around the same training times mentioned in the repository)

Can you tell me your requirement.txt

from jax3d.