0.6.0b6drm

0.6.0b6drm (Beta)

• Supported algorithms:
  • Random Forest
    • Regression
    • Classification (Binomial and Multinomial)
  • XGBoost
    • Regression
    • Classification (Binomial and Multinomial on selected Alveo cards)
  • LightGBM
    • Regression
    • Classification (Binomial and Multinomial on selected Alveo cards)
• Supported frameworks:
  • scikit-learn
  • XGBboost
  • LightGBM
• No feature scaling required: float32-based tree traversal algorithm in FPGA
• Kernel optimized for large ensambles and RF classification (greater than hundreds of trees)
• Single model inference
• Latency optimized inference
• Python interface with non-blocking calls
• Inference server with a single FPGA card
• DRM-based licensing system