Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Using 'ner' task in pipeline with a non default model gives me entities as "LABEL-6" , "LABEL-8" instead of "I-ORG" and "I-LOC"

See original GitHub issue

model - dbmdz/bert-base-cased-finetuned-conll03-english

Language - english

The problem arises when using:

my own modified scripts

The tasks I am working on is:

an official GLUE/SQUaD task: ner

To reproduce

tokenizer = AutoTokenizer.from_pretrained("dbmdz/bert-base-cased-finetuned-conll03-english")
model = AutoModelForTokenClassification.from_pretrained("dbmdz/bert-base-cased-finetuned-conll03-english")
ner_task = pipeline(task='ner', model=model, tokenizer=tokenizer)
ner_task('Hugging Face is a French company based in New-York.')

Issue Analytics

State:
Created 3 years ago
Comments:8

Top GitHub Comments

1reaction

stefan-itcommented, Apr 29, 2020

Oh no, there’s something wrong with the label alignment (can also be seen in the config.json):

"id2label": {
    "0": "LABEL_0",
    "1": "LABEL_1",
    "2": "LABEL_2",
    "3": "LABEL_3",
    "4": "LABEL_4",
    "5": "LABEL_5",
    "6": "LABEL_6",
    "7": "LABEL_7",
    "8": "LABEL_8"
  },
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "LABEL_0": 0,
    "LABEL_1": 1,
    "LABEL_2": 2,
    "LABEL_3": 3,
    "LABEL_4": 4,
    "LABEL_5": 5,
    "LABEL_6": 6,
    "LABEL_7": 7,
    "LABEL_8": 8
  },

Thanks for reporting, I’ll try to fix it 😃

In the meantime, you could use our large model dbmdz/bert-large-cased-finetuned-conll03-english 😅

0reactions

stale[bot]commented, Nov 5, 2020

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.