Comments (4)
@EdisonIsCoding As shown in https://treelite.readthedocs.io/en/latest/treelite-runtime-api.html#treelite.runtime.Predictor.predict, TreelitePredictorPredictBatch function is currently not thread safe. The reason is that it uses a dedicated thread pool to parallelize batch prediction. Consider using TreelitePredictorPredictInst instead.
from treelite.
Hi,
I'm developing a service recently which will allow others to do prediction via my service. And internally, I'm using a treelite predictor and calling into TreelitePredictorPredictBatch to make a prediction. So I'm wandering that whether this API is thread safe, since the request will come from different threads? Thanks for any help in advance!!
from treelite.
@hcho3 Thanks for your timely response! But in the case when the batch only have one row, it looks like TreelitePredictorPredictBatch also works fine, since I ran a stress test with 30 threads and each thread called TreelitePredictorPredictBatch API ~3k times with only one row in the batch (where I should call TreelitePredictorPredictInst in stead), the server did get the right prediction and didn't get into any problem. Is that true?
from treelite.
I'll have to come back to the stability of batch prediction API. I suppose that TreelitePredictorPredictBatch can work in your use case, but I won't provide any guarantees.
from treelite.
Related Issues (20)
- Support boosting from the average in sckit-learn
- [FEA]Add max_memory_usage param in model.export_lib
- Document how to use import_from_json()
- Release version 3.2.0
- How can I complie lgb model in msvc?
- Update task_type enum to distinguish between regressors and binary classifiers
- [Doc] Write a tutorial for integrating Treelite in a C++ project HOT 1
- Tracking issue for Treelite 3.9 and 4.0
- [Doc] Document Treelite serialization format. HOT 1
- Add first-class support for multi-output models in Treelite 4.0
- API refactors
- Documentation for library writers HOT 1
- treelite prediction 4x slower than xgboost HOT 3
- Document that Left child is chosen when condition is evaluated True
- Use std::variant to implement type-based dispatching
- Do not call np.squeeze on output of predict_leaf() / predict_per_tree() HOT 1
- treelite::ConcatenateModelObjects() ought to set threshold_type and leaf_output_type fields
- Clean up serialization logic
- Support XGBoost gblinear Booster HOT 1
- Release version 3.3.0
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 treelite.