Initial setup error(s)

[navaneetham-aicomputing]

@tscholak and team, thanks a lot for let PICARD available for public. I am sure this help many NLP researchers life easy. I did many experiments with PICARD(HF environment) works awesome. I try to explore PICARD in two of my local environments:

Environments Configurations:

1. MAC M1 chip (aware there are issues with few python packages)
2. GCP E2-Standard "Linux box with 32GB RAM & 8vCPU").
3. Python 3.9 was installed on both machines.

Followed below steps as per instructions given in README and issue(#36)

- git clone git@github.com:ElementAI/picard.git
- cd picard
- git submodule update --init --recursive
- poetry shell
- poetry install 

installation throw below error on both environments:

  **ValueError**
  picard is not a package.

Note: Please find the screenshot below

I did notice others also reported similar issue #36 . As per @tscholak recommendation I did "poetry install --no-root"
No errors. But when I try to run python seq2seq/run_seq2seq.py getting below error. Any idea what needs to be done to resolve this issue?

(py) ➜  picard git:(main) ✗ python seq2seq/run_seq2seq.py
None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
Traceback (most recent call last):
  File "seq2seq/run_seq2seq.py", line 20, in <module>
    from transformers.models.auto import AutoConfig, AutoTokenizer, AutoModelForSeq2SeqLM
ImportError: cannot import name 'AutoModelForSeq2SeqLM' from 'transformers.models.auto' (/Users/nav/opt/anaconda3/envs/py37/lib/python3.7/site-packages/transformers/models/auto/__init__.py)

Looks like I am missing something? I am happy to provide more information if needed.

[tscholak]

Hi, the code in this repository is not compatible with non-x86 architectures, unfortunately.

[iMayK]

Hi, the error is also there for x86.

[sbajpai2]

Even I am facing the similar issue.
Is this resolved?