hi everyone <!here> , we launched a contest on 99 design to find a logo for BentoML project. There are over 70 submissions so far, thought you guys might be interested to see this. Let us know if there’s one in the list that you really like! https://99designs.com/contests/925243/entries

Come check out our logo finalists

Vote which one you like!

Sorry. This is the poll address.

Hello

Hi 👋

I have deployed tabular model using bentoml

when I am sending the Curl I am getting Internal server error

Value Error : If using all scalar value you must pass an index

I have my notebook in Azure Machine and I am running it locally there

Ok. You trained the model in notebook and running it locally on the azure machine. Is that correct? Could you tell me what command you used for deploying? And after you packaged the bento bundle, did you import it and run it successfully against test data?

I followed the tabular example

of fastai

It maybe because the http request input is not in the right format, could you share the curl command?

And you are using DataframeHandler right?

curl -X POST http://127.0.0.1:5000/predict \ -H 'Content-Type: application/json' \ -d '[{ "ID": "AL011980", "Name": "UNNAMED", "Status": "TD", "Latitude": 29.8, "Longitude": -80.2, "Maximum.Wind": 25, "date_time": "1980-07-18 000000", "diff": 5, "rapid_int": 0, "i": 5, "n": 17, "persistence": 0, "product": 0, "Initial.Max": 25, "speed": 17.0712, "speed_z": 0.0351582, "speed_m": 9.26624, "Jday": 54, "Maximum.Wind_p": 25, "Latitude_p": 29.9, "Longitude_p": -79.3 }]'

I am using this curl command

I am running it on different data set

I haven't change any thing in the dataframe handler part

I am able to successfully install the tabular model

I can hit localhost in the browser and see its running on port 5000

@Dheeraj Gupta could you also share your “predict” function code?

This message was deleted.

hi everyone, I just found out about bentoML. is it possible to perform training as well as predict?

hi @tony hung, do you mean perform training as the API server receive a prediction request? Or you are looking for something to help with structure training code that integrates with BentoML?

the ideal scenario is a way to train the model at any given time. Im currently working on a fastai model and i need a way to retrain the model at any point in time.Say if there’s a new training example in a S3 bucket, i’d like to retrain with that new training example

i tried using Sagemaker but I had errors when trying to install dependencies (librosa)

Got it, sounds like you would like to be able to reload a new model as the API Server is running in production?