Mark Saroufim

@pmeier to clarify by backend I mean one of these https://github.com/pytorch/vision#image-backend - i.e: pillow, accimage, pillow simd etc.. Overall the new interface for adding datasets looks good but I'm more...

Seems like there's a good workaround too https://github.com/NVIDIA/DALI/issues/3081#issuecomment-866239816 - I'll take a more thorough look

I think so? Might be worth running a quick check with a very tiny value of max response size to see if you can reliably trigger the warning

@dalvlv typically dynamic input sizes are not supported by PyTorch natively as in you need to pad your inputs to some size which is what we do for example in...

Hi @hgong-snap is there any chance you can share a repro? In particular your `.mar` file and `config.properties` - otherwise I'll see if this occurs on smaller toy models

An update I tried to repro this offline ## Setup ``` git clone https://github.com/pytorch/serve mkdir model_store torchserve --start --model-store model_store python -m grpc_tools.protoc --proto_path=frontend/server/src/main/resources/proto/ --python_out=ts_scripts --grpc_python_out=ts_scripts frontend/server/src/main/resources/proto/inference.proto frontend/server/src/main/resources/proto/management.proto ``` ##...

@hgong-snap confirming we can finally repro, will get back to you with a solution soon

Yep, don't forget to run `ts_scripts/install_dependencies` first though with dev dependencies and GPU if needed

I agree! If you'd like to make a PR with better names I'd be happy to merge it

@sreenath-tm 1. Run this example https://github.com/pytorch/serve/tree/master/examples/Huggingface_Transformers and understand what the outputted underlying labels actually mean 2. Think of better names to give to the labels instead of Accepted vs Not...