Sentence-transformers: public.ukp.informatik.tu-darmstadt.de Unreachable

Created on 2 Sep 2020  路  20Comments  路  Source: UKPLab/sentence-transformers

It looks like the server which hosts the pre-trained models (https://public.ukp.informatik.tu-darmstadt.de/reimers/sentence-transformers/v0.2/) has been unavailable for a few hours now.

Most helpful comment

All 20 comments

Yes, there is currently a power outage which should be fixed soon.

Several pre trained models can also be found here:
https://huggingface.co/sentence-transformers

Looking forward when the server up again

Can you confirm how much time it will take to repair the power outage?

sys-admins sadly mention that this might take longer, as it looks like that the power adapter needs to be replaced.

I currently upload the models here:
https://sbert.net/models/

You can download them from there, unzip and use it like this

from sentence_transformers import SentenceTransformer
model = SentenceTransformer('path/to/unzipped/distilbert-base-nli-mean-tokens')

sys-admins sadly mention that this might take longer, as it looks like that the power adapter needs to be replaced.

I currently upload the models here:
https://sbert.net/models/

You can download them from there, unzip and use it like this

from sentence_transformers import SentenceTransformer
model = SentenceTransformer('path/to/unzipped/distilbert-base-nli-mean-tokens')

Thanks! It is working.

I just pushed a new version (0.3.5.1) that uses the new download server.

It can be installed via:

pip install -U sentence-transformers

Not all models are up yet, but I hope the most important models are already online.

which models are available? seem to be getting error when i tried a couple. also i get permission denied when i try to download the zip

Yes, there is currently a power outage which should be fixed soon.

Several pre trained models can also be found here:
https://huggingface.co/sentence-transformers

And here: https://huggingface.co/T-Systems-onsite/bert-german-dbmdz-uncased-sentence-stsb ;-)

which models are available? seem to be getting error when i tried a couple. also i get permission denied when i try to download the zip

Thanks for pointing this out. Some files did not have the right permission. I fixed that and now all the uploaded zips should be accessible.

Hi @nreimers

Coud you, please, upload the model _distiluse-base-multilingual-cased_ too?

Thanks in advance!

Ditto @georgenv - I depend on distiluse-base-multilingual-cased and our deployment is currently broken

Same here /reimers/sentence-transformers/v0.2/bert-base-nli-stsb-mean-tokens.zip

HI @nreimers
PLEASE upload distiluse-base-multilingual-cased model too.

Hi @ankitkr3 @DGaffney @georgenv

distiluse-base-multilingual-cased is now uploaded and can be used as before.

Hi, I am not able to download the datasets for running training_nli. I was wondering if you could upload the datasets as well?

Previous links to the dataset:
https://public.ukp.informatik.tu-darmstadt.de/reimers/sentence-transformers/datasets/AllNLI.tsv.gz & https://public.ukp.informatik.tu-darmstadt.de/reimers/sentence-transformers/datasets/stsbenchmark.tsv.gz

Thanks a lot!

Hi @nandinib1999
I am currently uploading them here:
https://sbert.net/datasets/

@nreimers thanks for uploading these models, can you please help with the pretrained roberta-large-nli-stsb-mean-tokens model also, at the huggingface it says deprecated. Also can you give an update when the power outage would be resolved.
Thanks in advance.

Hi @ankitkr3

I uploaded the two roberta models to sbert.net/models.

The power outage might take a bit longer. It appears that the server power supply was damaged during the outage and our system admins might need to get a new power supply for that server before it can come back online :/

Glad to see you put this to bed, @nreimers - must have been not too fun! Thanks for being in touch about it throughout.

Was this page helpful?
0 / 5 - 0 ratings

Related issues

lagatorc picture lagatorc  路  4Comments

zishanmuzeeb picture zishanmuzeeb  路  4Comments

hetianbaicnn picture hetianbaicnn  路  5Comments

FrancescoSaverioZuppichini picture FrancescoSaverioZuppichini  路  3Comments

RuskinManku picture RuskinManku  路  5Comments