Can I get the environment.yaml file?

[yujiapingff]

The paper of this work is impressive, and I am trying to follow your work and figure out that the environment.yaml file is missing. So can I get the environment.yaml to generate a suitable environment to run this project?

[laitifranz]

I recreated the env using pip, for those who need it. Note that if you want to use the accimage package you need conda, but they provide different implementations for the image loader, so you can safely use pip (see here).

requirements.txt file using Python 3.9 (Lightning is not needed)

[YCAyca]

The requirements.txt seems many stuff that is not used and it gives me package not found error while pip installing in Docker. I share the files working for me with Docker:

requirements.txt

uvicorn==0.17.4 loguru gunicorn ftfy regex scikit-learn timm matplotlib torch==1.13.1 torchaudio==0.13.1 torchmetrics==0.11.0 torchvision==0.14.1

Dockerfile

#FROM continuumio/miniconda3 ARG CUDA="10.1" ARG CUDNN="7" ARG PYTORCH="1.5"

FROM pytorch/pytorch:${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-runtime

WORKDIR sprompt

COPY requirements.txt requirements.txt

ENV PYTHONPATH "${PYTHONPATH}:/workspace/sprompt/models"

[4vicii]

hi, did you change cddb_sip.yaml net_type: slip to sip

i try to reproduce the vit-base method,but meet a bug:

pretrained_cfg = resolve_pretrained_cfg(variant, kwargs=kwargs) TypeError: resolve_pretrained_cfg() got an unexpected keyword argument 'kwargs'

my env is same to yours.

[YCAyca]

Hello, it gives me some other errors with sip and I didn't spend time to fix since I realized S-prompt is domain incremental and I need class incremental solution