Aradhye Agarwal comments

Results 9 comments of


                                            Aradhye Agarwal

Question about the reproducation of the results in the math_10k

@Zhenyu001225 any idea why this happens? An extreme case is for transformers 4.40.0 which gave me gibberish output as mentioned in [this](https://github.com/AGI-Edgerunners/LLM-Adapters/issues/63) issue. Thanks

Training speed

Hi @yongmayer, thanks for appreciating our work. So we used a per device batch size of 4 resulting in a total batch size of 32 with 8 GPUs. The speed...

Training speed

Hi @yongmayer. While we do use the stable diffusion backbone, we do not use it as a diffusion model per se. Instead we use the UNet (or backbone) as a...

Comparison with Depth Anything

Stay tuned!

Training reproduction problems

Please take a look at https://github.com/Aradhye2002/EcoDepth/issues/25.

CUDA OUT OF MEMORY during training

Hi @PreyumKr, Might I ask what per device batch size you used for training? Also, how much memory does (do) your GPU(s) have? As per our experiments, for the smallest...

Inference with outdoor model

Hi, Can you try visualizing the images? My suspicion is that the scale factor is incorrect since the a{1,2,3} values are coming 0. If this is indeed the case, then...

Regarding training a new model from scratch without using a trained pre-trained model

Hi @wanshishuns. We're delighted that you found our work helpful! The VIT_MODEL which we have defined in EcoDepth/model.py `google/vit-base-patch16-224`, is built for the ImageNet dataset. The ImageNet dataset has 1000...

onnx / TensorRT IN C++

I think this should be possible now that we have integrated EcoDepth with pytorch-lightning. We'll get back to you on this.