Comments (3)
Thanks for you information. Since we have added more codes after the FTT-NAS work, It seems like a bug introduced in our later up modification of the shared component classes (the evaluator, and weights manager). I'll check it as soon as possible.
from aw_nas.
I have fixed this bug in the latest commit a3834db.
The bug comes from: (If you are interested): In the early code, we aggregate the rewards/perfs. of each batch by mean
. Later on, when we are implementing Detection NAS, when implementing mAP reward calculation, this behavior is no longer suitable. Thus, while still keep mean
as the default, we permit the objective to provide special handling of the aggregate function of batch rewards. Previously, we check if the length of criterions
and number of perfs (len(aggregate_fns)
) is the same. However, the third criterion function return multiple perfs! After the results are flattened, the length of criterion results should match the number of perfs.
from aw_nas.
Thank you for your careful hints and explanations. I will continue to run the code according to your instructions. Thank you very much and wish you all the best.
from aw_nas.
Related Issues (20)
- Hardware: Profiling Network Generation Problem
- About FTT-NAS GPU search problem... HOT 3
- I seem to found a small bug in your FTT-NAS project... HOT 4
- ray_dispatcher
- ray version HOT 1
- Adversarial Supernet training HOT 5
- Similar DAG encoding scheme to an existing paper HOT 1
- Generate sample config error and FBNet support? HOT 1
- nasbench.api.OutOfDomainError: unsupported op none HOT 1
- The program is stuck in an endless loop, when run awnas search examples/nasbench/nasbench-101_gates_sa.yaml HOT 1
- How to tune each rollout and how to use the Pareto-based EA HOT 2
- In aw_nas/aw_nas/objective/zerocost.py, the module foresight cannot be imported. HOT 2
- How to change the environment variable “AWNAS_HOME“ HOT 2
- Missing data dumping code for NAS-Bench-201 in gates. HOT 2
- Question about the supernet training?
- question about supernet training.
- Code for TA-GATES HOT 3
- Regrading TA-GATES dataset HOT 1
- ENAS, NB301 dataset preprocessing code for TA-GATES HOT 6
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 aw_nas.