Skip to content

README.md

Latest commit

 

History

History
72 lines (55 loc) · 1.74 KB

README.md

File metadata and controls

72 lines (55 loc) · 1.74 KB

FLIP Inference

Inference pipeline for FLIP models, providing ONNX export and WebAssembly compilation for efficient deployment.

Features

  • ONNX Export: Convert trained PyTorch models to ONNX format with KV caching optimization
  • WebAssembly Support: Compile C extensions to WASM for browser-based inference
  • Optimized C Extensions: High-performance patch sampling and Gaussian operations
  • Evaluation Tools: Benchmark model performance on HDF5 datasets

Setup

1. Create Environment

conda env create -f environment.yml
conda activate flip-inference

Note: Default environment includes GPU support. For CPU-only, modify environment.yml:

  • Replace pytorch with pytorch-cpu
  • Replace onnxruntime-gpu with onnxruntime

2. Build C Extensions

cd ext
python setup.py build install
cd ..

3. (Optional) WebAssembly Compilation

For browser deployment, first install Emscripten:

# Install Emscripten
git clone https://github.com/emscripten-core/emsdk.git
cd emsdk
./emsdk install latest
./emsdk activate latest
source ./emsdk_env.sh
cd ..

# Compile to WebAssembly
cd ext
chmod +x compile_wasm.sh
./compile_wasm.sh
cd ..

Quick Start

  1. Export trained model to ONNX:

    python export_to_onnx.py \
    --config configs/flip-large.json \
    --checkpoint chceckpoints/flip-large.ckpt \
    --encoder-output flip-encoder-large.onnx \
    --predictor-output flip-predictor-large.onnx

  2. Run evaluation:

    python -m model.scripts.evaluate_single_hdf5_onnx \
    --dataset_path your_dataset.hdf5 \
    --encoder_path flip-encoder-large.onnx \
    --predictor_path flip-predictor-large.onnx \
    --config configs/flip-large.json \