No way to load SequenceTagger

As it was proposed by Using Ancient Greek Notebook.
tagger = SequenceTagger.load(‘SuperPeitho-FLAIR-v2/’)
leads to
HTTPError: 401 Client Error: Unauthorized for url:

Neither it is possible to load it from locally lfs downloaded

Hi @Bonum, the notebook has following example:

from transformers import AutoTokenizer, AutoModel
tokeniser = AutoTokenizer.from_pretrained("pranaydeeps/Ancient-Greek-BERT")
model = AutoModel.from_pretrained("pranaydeeps/Ancient-Greek-BERT")  

These models are hosted on HF due to which will work directly.

While you are trying to use SuperPeitho-FLAIR-v2/ which isn’t, as the author told in the README.

We provide a pre-trained POS Tagging model in the directory SuperPeitho-FLAIR-v2

That means, you need to clone the main repository and use the model file directly.

$> git clone
$> cd Ancient-Greek-BERT

## run your python code along with above load command

All the best ! :wave:

Hi @shivam. Thank you for your reply!
Indeed after realising that SuperPeitho-FLAIR-v2/ is not hosted on HF I cloned the git repo, downloaded the lfs .pt file, installed flair framework and tried to load it locally:
tagger = SequenceTagger.load(’/content/ag/SuperPeitho-FLAIR-v2/’)
I don’t have GPU resources, so i work on CoLab. From error logs :
‘…tokenizer_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)’
it is seen that the tagger tries to load some configs which don’t exist in
Although the author claims that it is a pre-trained POS Tagging model with FLAIR framework, aparently the config files are not included in it.
It would be great if we could use FLAIR framework, which has powerful NER functions for morphological analysis.

Hi @Bonum, here is an working example [should work on Colab] on how to use SequenceTagger:

Yup, the files are missing in GitHub repository, while available in HF, so I have just merged them both to make it work. :man_shrugging:

All the best !

Hi @shivam

Thank you very much for your workaround of the problem!
Your example notebook works fine on colab.

Best regards!