Project was developed on mac os, tested on linux x86 Used python 3.9.1
Contest: https://contest.com/docs/ML-Competition-2023
Contest first results: https://t.me/contest/341
This project line:
| Entry | Result | Time
----+-------+--------+-------
6. | 4672 | 74.5% | 265ms
cd src
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
cmake --build .
- Run
dataset-prepare.ipynb
to prepare dataset fromraw_data_v2.zip
. In result: 50000+ batches indataset_v4o_train
and 5000 batches indataset_v4o_validate
- Run
dataset_tg-prepare.ipynb
to prepare dataset fromml2023-r1-dataset.tar
andml2023-d1-dataset.tar
- Run
train.ipynb
to train on datadataset_v4o
batches. And save onnx model - Watch at
convert-model.ipynb
to convert onnx model to ort and compile onnxruntime