train config yaml system does not correctly load local variable

[joihn]

hanks for this repo,
I'm launching a training with python 3.10, via pycharm IDE.
when executing:
python train.py --config_file=./config/image_gen_train.yaml

the nominal user would expect image_gen_train.yaml params to be loaded.
However they are NOT.

diffinfinite/train.py

Line 34 in 4d23096

locals().update(config[key])

this syntax seems uneffective to change local variable. Thus local variable stay unchanged by the config.

relevant stackoverflow :
https://stackoverflow.com/questions/30295163/update-locals-from-inside-a-function

env:
ubuntu 20 with the below presented docker container
python 3.10.11
dockerfile for my developer environment:

# Dockerfile inspired by https://github.com/open-mmlab/mmsegmentation/blob/1.x/docker/Dockerfile

ARG PYTORCH="2.0.1"
ARG CUDA="11.7"
ARG CUDNN="8"

ARG PYTHON_VERSION="3.10"

FROM pytorch/pytorch:${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel as diffusion
ARG DEBIAN_FRONTEND=noninteractive

ENV TORCH_CUDA_ARCH_LIST="6.0 6.1 7.0 7.5 8.0 8.6 8.9+PTX"
ENV TORCH_NVCC_FLAGS="-Xfatbin -compress-all"
ENV CMAKE_PREFIX_PATH="$(dirname $(which conda))/../"

RUN apt-get update && apt-get install -y git ninja-build libglib2.0-0 libsm6 libxrender-dev libxext6 libgl1 wget vim nano \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*

COPY requirements.txt /tmp/requirements.txt
RUN pip install -r /tmp/requirements.txt
RUN pip install scikit-learn opencv-python pandas

[joihn]

I can confirm that launching a training via CLI (not via pycharm ) results in the same behavior

[marcoaversa]

thanks for your feedback, the code will be fixed as soon as possible.