Maxim
@backpropaganda.bsky.social
Doing Computer Vision stuff with ML.
This is just your standard gradient accumulation?
for i, (input, target) in enumerate(data):
output = model(input)
loss = loss_fn(output, target)
loss = loss / iters_to_accumulate
loss.backward()
if (i + 1) % iters_to_accumulate == 0:
optimizer.zero_grad()
for i, (input, target) in enumerate(data):
output = model(input)
loss = loss_fn(output, target)
loss = loss / iters_to_accumulate
loss.backward()
if (i + 1) % iters_to_accumulate == 0:
optimizer.zero_grad()
December 19, 2024 at 7:02 PM
This is just your standard gradient accumulation?
for i, (input, target) in enumerate(data):
output = model(input)
loss = loss_fn(output, target)
loss = loss / iters_to_accumulate
loss.backward()
if (i + 1) % iters_to_accumulate == 0:
optimizer.zero_grad()
for i, (input, target) in enumerate(data):
output = model(input)
loss = loss_fn(output, target)
loss = loss / iters_to_accumulate
loss.backward()
if (i + 1) % iters_to_accumulate == 0:
optimizer.zero_grad()
Still a bit confused on when to use Illuminate vs NotebookLM for getting an audio overview of papers. Currently using Illuminate.
December 14, 2024 at 10:51 PM
Still a bit confused on when to use Illuminate vs NotebookLM for getting an audio overview of papers. Currently using Illuminate.
I find 'uv sync' so fast that I would just change the version in project.toml and .python-version and sync again. The fact that the env is tied to a directory may sometimes be a negative but it also ensures all package versions are tracked in git.
December 4, 2024 at 6:27 AM
I find 'uv sync' so fast that I would just change the version in project.toml and .python-version and sync again. The fact that the env is tied to a directory may sometimes be a negative but it also ensures all package versions are tracked in git.
What seems to be currently the best approach for depth estimation, diffusion models or "old-school" discriminative models? Both seem to claim SOTA models nowadays?
December 2, 2024 at 2:33 PM
What seems to be currently the best approach for depth estimation, diffusion models or "old-school" discriminative models? Both seem to claim SOTA models nowadays?
Every once in a re-read Joseph Redmon's YOLOv3 paper. That was really a work of art..
"Sometimes you just kinda phone it in for a year, you know? I didn’t do a whole lot of research this year. [...] I managed to make some improvements to YOLO. But, honestly, nothing like super interesting"
"Sometimes you just kinda phone it in for a year, you know? I didn’t do a whole lot of research this year. [...] I managed to make some improvements to YOLO. But, honestly, nothing like super interesting"
November 29, 2024 at 3:39 PM
Every once in a re-read Joseph Redmon's YOLOv3 paper. That was really a work of art..
"Sometimes you just kinda phone it in for a year, you know? I didn’t do a whole lot of research this year. [...] I managed to make some improvements to YOLO. But, honestly, nothing like super interesting"
"Sometimes you just kinda phone it in for a year, you know? I didn’t do a whole lot of research this year. [...] I managed to make some improvements to YOLO. But, honestly, nothing like super interesting"
Would be interesting to see how it would perform for BOP dynamic onboarding.
As you can tell, I’ve started sharing interesting 6D pose estimation papers I come across. I already track these for myself, so why not share them with all of you?
As you can tell, I’ve started sharing interesting 6D pose estimation papers I come across. I already track these for myself, so why not share them with all of you?
November 26, 2024 at 2:02 PM
Would be interesting to see how it would perform for BOP dynamic onboarding.
As you can tell, I’ve started sharing interesting 6D pose estimation papers I come across. I already track these for myself, so why not share them with all of you?
As you can tell, I’ve started sharing interesting 6D pose estimation papers I come across. I already track these for myself, so why not share them with all of you?
Authors: Vincent van der Brugge, Marc Pollefeys, Joshua B. Tenenbaum, Ayush Tewari, Krishna Murthy Jatavallabhula
Arxiv: arxiv.org/abs/2411.1...
Code: github.com/vincentva...
Arxiv: arxiv.org/abs/2411.1...
Code: github.com/vincentva...
PickScan: Object discovery and reconstruction from handheld interactions
Reconstructing compositional 3D representations of scenes, where each object is represented with its own 3D model, is a highly desirable capability in robotics and augmented reality. However, most...
arxiv.org
November 26, 2024 at 2:02 PM
Authors: Vincent van der Brugge, Marc Pollefeys, Joshua B. Tenenbaum, Ayush Tewari, Krishna Murthy Jatavallabhula
Arxiv: arxiv.org/abs/2411.1...
Code: github.com/vincentva...
Arxiv: arxiv.org/abs/2411.1...
Code: github.com/vincentva...
As you can tell, I’ve started sharing interesting 6D pose estimation papers I come across. I already track these for myself, so why not share them with all of you?
November 25, 2024 at 2:05 PM
As you can tell, I’ve started sharing interesting 6D pose estimation papers I come across. I already track these for myself, so why not share them with all of you?
Authors: Kai Chen, Yiyao Ma, Xingyu Lin , Stephen James, Jianshu Zhou, Yun-Hui Liu, Pieter Abbeel, Qi Dou
Openreview: https://openreview.net/forum?id=FTpKGuxEfy
Project page: https://vfm-6d.github.io/
Openreview: https://openreview.net/forum?id=FTpKGuxEfy
Project page: https://vfm-6d.github.io/
Vision Foundation Model Enables Generalizable Object Pose Estimation
Object pose estimation plays a crucial role in robotic manipulation, however, its practical applicability still suffers from limited generalizability. This paper addresses the challenge of...
openreview.net
November 25, 2024 at 2:05 PM
Authors: Kai Chen, Yiyao Ma, Xingyu Lin , Stephen James, Jianshu Zhou, Yun-Hui Liu, Pieter Abbeel, Qi Dou
Openreview: https://openreview.net/forum?id=FTpKGuxEfy
Project page: https://vfm-6d.github.io/
Openreview: https://openreview.net/forum?id=FTpKGuxEfy
Project page: https://vfm-6d.github.io/
The GPU poor do not have it easy ;). But usually it is just multiple notebooks and nvidia-smi is handy to see how much each notebook is taking up.
November 21, 2024 at 3:04 PM
The GPU poor do not have it easy ;). But usually it is just multiple notebooks and nvidia-smi is handy to see how much each notebook is taking up.
Haven't switched yet, is there an easy way to see which programs take up how much gpu memory like in nvidia-smi?
November 21, 2024 at 9:16 AM
Haven't switched yet, is there an easy way to see which programs take up how much gpu memory like in nvidia-smi?
Thats why all my pytorch code looks like:
```
from torchvision.transforms.v2.functional import to_dtype, to_image
img_tensor = to_dtype(to_image(image), scale=True)
```
```
from torchvision.transforms.v2.functional import to_dtype, to_image
img_tensor = to_dtype(to_image(image), scale=True)
```
November 20, 2024 at 7:19 PM
Thats why all my pytorch code looks like:
```
from torchvision.transforms.v2.functional import to_dtype, to_image
img_tensor = to_dtype(to_image(image), scale=True)
```
```
from torchvision.transforms.v2.functional import to_dtype, to_image
img_tensor = to_dtype(to_image(image), scale=True)
```