Hi Swapna, Thanks for the FLIP and sorry for the late reply. After reading it, I have some comments about the proposal.
1. I notice that the parameter `model-directory-path` is introduced. It should work but I see that some popular python model libraries like `transformer` and `vllm`, tend to directly use the parameter name `model` which could be a local path or a simple name like `Qwen/Qwen3-0.6B`. Then they check if `model` is a path or a simple name and automatically download it to local cache directory if necessary. Considering use cases based these libraries, maybe `model-directory-path` can be renamed to `model` which can be either a real path or a simple model name. 2. There is a `set_model_config` in the python PredictFunction interface. I find that in the previous discussion, you are considering passing PredictRuntimeContext in `open`. I do agree with this choice as well, which makes the logic more clear and uses can just use the configs in the `open` method to init their model. I just find that the FLIP still has the `set_model_config` so just want to double check it here. Thanks for your time! Best, Biao Geng > 2025年9月22日 23:36，Swapna Marru <[email protected]> 写道： > > Hi Devs, > > > I am interested in learning more about MODEL, ML_PREDICT, and ML_EVALUATE > functionalities added in the following FLIP. > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-437%3A+Support+ML+Models+in+Flink+SQL > > > I see the original FLIP has extensibility to local model providers in Flink. > > > Is there a way to do pluggable local model providers in Python? Like, say, > generate embeddings using Sentence transformer models running locally in > Flink. > > > An option could be to introduce a Model Provider factory implementation in > Java that internally uses a predict function in Python . But I see this > puts in a lot of work related to Java to Python communication/translation > inside the provider. > > > Something like PythonRuntimeProvider along with PredictRuntimeProvider / > AsyncRuntimeProvider which can handle Java -> Python translations out of > the box would be helpful to de-duplicate that effort. > > > Can you please point to, if there are any discussions related to this > already ? Or any other ways to achieve the same? Please share your thoughts. > > > -Thanks, > > Swapna Marru
