Comments (2)
Sure thing! You don't want training weights to be low-precision, or it can cause instability. Therefore, when not using LoRA, we want to keep the UNet params in full float32. If we're using LoRA, we're not training the UNet params, so we can cast them to low precision. That's why the variable is called inference_dtype
.
from ddpo-pytorch.
I see, thanks!
from ddpo-pytorch.
Related Issues (20)
- Suggest to use larger gradient accumulation steps instead of multi GPUs HOT 3
- About the training with prompt_image_alignment configuration which uses llava_bertscore reward function HOT 5
- Question about the optimized objective. HOT 1
- Hello, when I trained an aesthetic model using the default configuration on 8 A800 cards, I found that the training process got stuck after completing one epoch, but it worked fine when using a single A800 card. May I ask what could be the cause of this situation? HOT 8
- Support for other schedulers HOT 1
- prompt-dependent value function optimization
- SDXL Support? HOT 1
- unet keeps producing nan during training HOT 2
- Code logics, thanks
- On reproducibility and LoRA HOT 1
- Batch size unrecogonized HOT 1
- Finetuning on google colab
- Questions about the reward curve and bert. HOT 1
- OOM when using "stabilityai/stable-diffusion-2-1" with batch size of 2 HOT 1
- reproducing the aesthetic experiment HOT 7
- OOM despite using A100-80GB GPU and 128GB CPU memory (+16 CPUs per task) HOT 2
- On reproducing LLaVA alignment experiments. HOT 2
- Prompt Alignment with LLaVA-server: Client-side prompt and image doesn't match server side reward HOT 4
- About the large dataset and Unet Training HOT 3
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 ddpo-pytorch.