SciSharp STACK
@ScisharpStack
A .NET based Open Source Ecosystem for Data Science, Machine Learning and AI.
You might like
A blog about how to connect LLamaSharp and BotSharp UI: medium.com/scisharp/build…
Host a LLM Powered Chatbot in .NET haiping008.medium.com/host-a-llm-pow…
Just released BotSharp v0.9.x github.com/SciSharp/BotSh…, introduced two main features: Routing and Templating to make your AI Bot journey be more smooth. Routing is for managing mulitple agents extremly easier. Templating is for making dynamic prompt with flexibility.
Train your deep learning model without leaving .NET. Released a new version to support Time Series in RNN.
Check out if you can run though by this updated Get Started, The BotSharp can work with LLMs and Vector Stores now. botsharp.readthedocs.io/en/latest/inst…
Another milestone is getting close. This PR makes .NET be able to load TensorFlow Keras SavedModel. That means we can use the state-of-art advanced model in .NET. github.com/SciSharp/Tenso…
A big PR to support Keras to saved model in .NET and load from Python to train. It looks good and merge soon.
After a year of hard work, our book is finally published, and if you want to support the sustainable development of software, you can buy genuine books from here: item.jd.com/13441549.html
Just released TensorFlow v0.70.0 and Keras .NET to support TimeSeries model. Check the example repo if you're interested in predicting feature(s) along with time axis.
No memory leak finally for Object Detection model training in TensorFlow .NET. The memory usage keeps stable about 1.6g~5g.
TensorFlow NumPy will be introduced in tf .net v0.60. The full functionality will be supported as time, most excited thing is the NDArray is allocated and computed on your GPU. Check it out! "using Tensorflow.NumPy;"
Released TensorFlow 2.5.0 CPU/ GPU native library in NuGet.
Recently I took time to review NumSharp code, I'm feeling it's time to refactor it again, currently it's not maintainable for contributors, difficult to fix bugs or add new features. The good news is we can make it better indeed, this is a roughly comparsion, it's 4x faster.
It's been merged finally, TString API will be available along with TensorFlow v2.5, that fixes the annoying breaking change related to string api.
Text preprocessing is a big part of machine learning, and we should have this feature.
United States Trends
- 1. #SwiftDay 8,130 posts
- 2. Columbus 63.5K posts
- 3. #WWERaw 18.8K posts
- 4. #IDontWantToOverreactBUT N/A
- 5. #TSTheErasTour 2,026 posts
- 6. $ZOOZ 1,019 posts
- 7. #IndigenousPeoplesDay 2,389 posts
- 8. Knesset 110K posts
- 9. Marc 35.4K posts
- 10. Good Monday 38.2K posts
- 11. Victory Monday 1,248 posts
- 12. Thanksgiving 42.6K posts
- 13. Flip 49.1K posts
- 14. Broadcom 1,397 posts
- 15. Penta 4,630 posts
- 16. Branch 47.4K posts
- 17. Rod Wave 2,941 posts
- 18. Kairi 9,673 posts
- 19. The Final Show 25.1K posts
- 20. Happy 250th 1,777 posts
Something went wrong.
Something went wrong.