Comments (6)
Thanks for your interest in our work. We do have a plan to release training and evaluation code in the future. However, the schedule is not clear yet. Please stay tuned for future updates.
#1 (comment)
from marigold.
I find it can be trained following the structure of the pipeline design, using the HuggingFace text_to_image.py
example Here, only changing the UNet part.
I have implemented an unofficial training version using the HuggingFace's diffusers when using the batch size equal 1 or 2, the training takes about 21G VRAM. Not sure whether can reproduce the result or not. Basically, the author says it can be trained on a single RTX4090 is True :)
from marigold.
Thank you for sharing this! I will try it at the earliest opportunity and will share any outcomes here.
from marigold.
I find it can be trained following the structure of the pipeline design, using the HuggingFace
text_to_image.py
example Here, only changing the UNet part.I have implemented an unofficial training version using the HuggingFace's diffusers when using the batch size equal 1 or 2, the training takes about 21G VRAM. Not sure whether can reproduce the result or not. Basically, the author says it can be trained on a single RTX4090 is True :)
Have you tried the "Annealed multi-resolution noise" mentioned in the paper? I'm not sure if I set it right in my training.
from marigold.
Ohh, about this point, I did not implement the Annealed multi-resolution noise......
from marigold.
I find it can be trained following the structure of the pipeline design, using the HuggingFace
text_to_image.py
example Here, only changing the UNet part.I have implemented an unofficial training version using the HuggingFace's diffusers when using the batch size equal 1 or 2, the training takes about 21G VRAM. Not sure whether can reproduce the result or not. Basically, the author says it can be trained on a single RTX4090 is True :)
Hello, I am sorry to bother you. Firstly, thanks a lot for writing the training pipeline code. I wonder whether you have trained your image2depth diffusion model based on the scripts. Did they perform good training results?
Thank you! :)
from marigold.
Related Issues (20)
- how to convert a HF Diffusers saved pipeline to a Stable Diffusion checkpoint? HOT 1
- the test results seem have much noise HOT 4
- Regarding the training convergence HOT 4
- Training on Custom Dataset HOT 1
- the prediction on in-the-wild example is noisy HOT 1
- Is code from the Bas-relief available
- Why set NaN depth values to zero on preprocessing? HOT 1
- Request for vkitti_val.tar and vkitti_vis.tar files HOT 1
- How to Manage the Large Hypersim Dataset for Reproduction? HOT 3
- ask for LCM distillation code HOT 2
- Low-Rank(LoRA) training of Marigold
- Multi-GPU Training HOT 1
- Clarification Needed: Training and Inference Pipeline HOT 4
- Any reason for not using vae.std to generate RGB latent? HOT 1
- do you plan to release better and more accurate models for this original marigold? HOT 1
- The purpose of using v_prediction as the target? HOT 1
- where can I get the "output/marigold_base/checkpoint/latest" HOT 1
- The demo 3D looks ok but not match with the predicted depth image HOT 1
- How to organize the vkitti data HOT 1
- train the model on my custom dataset HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from marigold.