Comments (7)
Thanks for your interest!
The random numbers are used in secret sharing to "hide" the plain number, which might cause the arithmetic overflow due to the summing up operation. Thus we confine the range of random numbers and apply the mod operation. The range can be adjusted according to the applications. And we encourage setting a large upper bound to ensure the privacy protection strength.
from federatedscope.
If you confine the range of the random numbers, then it is not safe any more, in the sense that the distribution of the random numbers is no longer uniform. Particularly, the random numbers should also take negative values (and then take mod operation).
from federatedscope.
Thanks for your suggestions! Maybe we can improve our implementation later.
BTW, such ''no longer uniform'' phenomenon is caused by the implementation of np.random.randint
?
from federatedscope.
Thanks for your suggestions! Maybe we can improve our implementation later. BTW, such ''no longer uniform'' phenomenon is caused by the implementation of
np.random.randint
?
Due to the confinement of "low=0".
For example, maybe use "np.random.randint(low=-self.mod_number, high=self.mod_number, size=shape)", and "np.random.randint(low=-2*self.mod_number, high=2*self.mod_number, size=shape)", etc.
from federatedscope.
Got it. I have misunderstood the ''confine'' (refer to the high=mod_number) before.
Thanks again for your comments :)
from federatedscope.
Got it. I have misunderstood the ''confine'' (refer to the high=mod_number) before. Thanks again for your comments :)
You are more than welcome.
For "high=self.mod_number", it is fine, since we need a "high end (range)" anyway in practical implementations.
from federatedscope.
都是国人,非要用英文来交流吗🤣
from federatedscope.
Related Issues (20)
- the issue of introducing a new package HOT 1
- Issue in DistributedRunner
- save client's model HOT 3
- In LLM, where is the adapter for clients and servers to interact? HOT 3
- Error when using register_data HOT 3
- Offsite tuning code with multigpu setting throws error HOT 3
- Smaller test/val loss but lower evaluation accuracy HOT 3
- How to use multi GPU to finetune Llama2 HOT 2
- Unable to run demo in hyperparameter optimization HOT 2
- Server global evaluation total number HOT 1
- Error with 4 bit quantized LLM HOT 3
- TypeError: call_file_data() missing 1 required positional argument: 'client_cfgs' HOT 6
- Some questions about Backdoor Bench HOT 2
- 训练得到的total_flops是负数 HOT 2
- LDA splitter:ValueError: too many values to unpack (expected 2) HOT 2
- how can i use my saving model HOT 1
- how can i use my saving model HOT 3
- How can i get the output result? HOT 1
- Soft prompt Tuning
- one bug in federatedscope/gfl/fedsageplus/trainer.py
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 federatedscope.