Ayush Chordia

Handler file(text_handler.py): ``` from abc import ABC import json import logging import os import ast import torch from transformers import MBartForConditionalGeneration, MBart50TokenizerFast from ts.torch_handler.base_handler import BaseHandler logger = logging.getLogger(__name__) class...

```Model server started. 2021-03-16 20:18:28,917 [INFO ] pool-2-thread-1 TS_METRICS - CPUUtilization.Percent:100.0|#Level:Host|#hostname:563d8708dab6,timestamp:1615925908 2021-03-16 20:18:28,918 [INFO ] pool-2-thread-1 TS_METRICS - DiskAvailable.Gigabytes:33.27151870727539|#Level:Host|#hostname:563d8708dab6,timestamp:1615925908 2021-03-16 20:18:28,919 [INFO ] pool-2-thread-1 TS_METRICS - DiskUsage.Gigabytes:18.78311538696289|#Level:Host|#hostname:563d8708dab6,timestamp:1615925908 2021-03-16 20:18:28,919 [INFO...

@dhanainme Just to clarify, all these issues only occur when trying to run the torchserve inside a docker image, there are no issues when running in a standalone ubuntu system....

@dhanainme I have tried everything possible to run this mar file inside a docker container using torchserve, every single time it fails with absolutenly no indication/logs of what's the underlying...

@maaquib Running the mar file on a standalone ubuntu box correctly, but the same mar file when mounted to torchserve docker image fails to load. When using the shell to...

@maaquib The underlying issue is the docker can't handle any custom handler, I literally wrote the entire docker image from scratch, it still failed with the following logs ``` 2021-03-22...

@maaquib Even mounting an existing mar file which runs on my ubuntu system to a docker image fails to execute on the image provided by pytorchserve ``` docker run -p...

@HenryDashwood @tobigue I am looking at your implementation of mbart on onnx and wanted to see if you had any feedback on how to make this work for mBart-50 many...