Comments (3)
Hi @thompsondd,
Could you please tell us which proxy you were using? Looks to me like removing an inplace relu operation somewhere in the Nasbench101 graph will fix the issue.
Thanks!
from naslib.
Thank you for your reply, @Neonkraft.
I am trying to use the Synflow proxy in NAS101 but the arch "(0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 2, 2, 4, 3, 3, 1)" raises this error (This is just one of the error cases that I have met).
Following the source code, I have removed the inplace of relu operation in https://github.com/automl/NASLib/blob/8c45f19dc259956c3bd253071135c798ad3df8ce/naslib/search_spaces/nasbench101/base_ops.py#L18C3-L19C10, but nothing changed.
Could you please tell me what you have modified the code?
from naslib.
Hi @thompsondd,
I have tried to reproduce your error and had no problem evaluating the zero-cost score for the architecture. Here's a snippet of code that I tried. You can correct me if it doesn't exactly match your case.
import logging
from naslib.predictors import ZeroCost
from naslib import utils
from naslib.utils import setup_logger, get_dataset_api
from naslib.search_spaces.nasbench101.conversions import convert_tuple_to_spec
from naslib.search_spaces import NasBench101SearchSpace
config = utils.get_config_from_args(config_type="zc")
logger = setup_logger(config.save + "/log.log")
logger.setLevel(logging.INFO)
utils.set_seed(config.seed)
utils.log_args(config)
dataset_api = get_dataset_api("nasbench101", config.dataset)
graph = NasBench101SearchSpace(n_classes=10)
test_arch = (0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0,
0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0,
0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1,
0, 0, 0, 0, 0, 0, 0, 0, 2, 2, 4, 3, 3, 1)
spec = convert_tuple_to_spec(test_arch)
graph.set_spec(spec)
predictor = ZeroCost(method_type="synflow")
train_loader, _, test_loader, _, _ = utils.get_train_val_loaders(config)
graph.parse()
score = predictor.query(graph, train_loader)
print("Zero cost score:", score)
logger.info('Test experiment complete.')
I had a synflow
score of 125.99:
Is it possible that I missed something or a version-related problem? Maybe you can also try running the same snippet, and tell what you are getting.
from naslib.
Related Issues (20)
- Drive links no longer exist HOT 1
- How can I use NASLib with my own data? HOT 1
- Train loader for NAS-Bench-ASR
- Checkpoint restart training HOT 2
- Search algorithms for zero-shot NAS
- Units tests failing on commits HOT 2
- Taskonomy dataset loading HOT 2
- Recreate a subset of benchmarks HOT 1
- How to initialize a NASBench-301 model for training? HOT 2
- This errors that occur when I validate the setup by running tests HOT 1
- Zero-Cost Metrics HOT 2
- TransNAS-Bench-101 Download HOT 2
- Extra .json files for zc_intro.py HOT 3
- How to get the trained weights of architecture network from NASbenchmark
- Issue with Running Predictor Experiments Due to Missing benchmarks Folder
- XGBoost experiments generalization HOT 2
- Downloading taskonomy with download_tnb.sh HOT 2
- The setup process doesn't work for me HOT 2
- Reasons not to use Optuna? HOT 2
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 naslib.