On https://github.com/SeldonIO/MLServer/blob/master/runtimes/xgboost/mlserver_xgboost/xgboost.py#L21 you can see that MLServer is looking for an XGBoost model file called "model.bst". However, I cannot find any reference to that file extension in the XGBoost documentation. As far as I can see, XGBoost's documented file extensions are:
- ".json" added in 1.0.0, an "open format that can be easily reused"
- ".ubj" for Universal Binary JSON format, available in 1.6.0
- ".model" for the "old binary internal format" prior to 1.0.0, as shown in examples
Where does MLServer get the ".bst" extension from, and what model format does it use? Shouldn't it use one of the extensions mentioned in the XGBoost documentation instead, to avoid ambiguity?
On https://github.com/SeldonIO/MLServer/blob/master/runtimes/xgboost/mlserver_xgboost/xgboost.py#L21 you can see that MLServer is looking for an XGBoost model file called "model.bst". However, I cannot find any reference to that file extension in the XGBoost documentation. As far as I can see, XGBoost's documented file extensions are:
Where does MLServer get the ".bst" extension from, and what model format does it use? Shouldn't it use one of the extensions mentioned in the XGBoost documentation instead, to avoid ambiguity?