README.md

---
license: apache-2.0
tags:
- token-classification
datasets:
- conll2003
metrics:
- precision
- recall
- f1
- accuracy

model-index:
- name: distilroberta-base-ner-conll2003
  results:
  - task:
      name: Token Classification
      type: token-classification
    dataset:
      name: conll2003
      type: conll2003
    metrics:
      - name: Precision
        type: precision
        value: 0.9492923423001218
      - name: Recall
        type: recall
        value: 0.9565545901020023
      - name: F1
        type: f1
        value: 0.9529096297690173
      - name: Accuracy
        type: accuracy
        value: 0.9883096560400111
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# distilroberta-base-ner-conll2003

This model is a fine-tuned version of [distilroberta-base](https://huggingface.co/distilroberta-base) on the conll2003 dataset.

eval F1-Score: **95,29** (CoNLL-03)   
test F1-Score: **90,74** (CoNLL-03)  

eval F1-Score: **95,29** (CoNLL++ / CoNLL-03 corrected)  
test F1-Score: **92,23** (CoNLL++ / CoNLL-03 corrected)  


## Model Usage

```python
from transformers import AutoTokenizer, AutoModelForTokenClassification
from transformers import pipeline

tokenizer = AutoTokenizer.from_pretrained("philschmid/distilroberta-base-ner-conll2003")
model = AutoModelForTokenClassification.from_pretrained("philschmid/distilroberta-base-ner-conll2003")

nlp = pipeline("ner", model=model, tokenizer=tokenizer,grouped_entities=True)
example = "My name is Philipp, I am a Machine Learning Engineer at HuggingFace and live in Nuremberg"

ner_results = nlp(example)
print(ner_results)
```


## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 4.9902376275441704e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 6.0
- mixed_precision_training: Native AMP

### Training results

#### CoNNL2003

It achieves the following results on the evaluation set:
- Loss: 0.0583
- Precision: 0.9493
- Recall: 0.9566
- F1: 0.9529
- Accuracy: 0.9883

It achieves the following results on the test set:
- Loss: 0.2025
- Precision: 0.8999
- Recall: 0.915
- F1: 0.9074
- Accuracy: 0.9741

#### CoNNL++ / CoNLL2003 corrected

It achieves the following results on the evaluation set:
- Loss: 0.0567
- Precision: 0.9493
- Recall: 0.9566
- F1: 0.9529
- Accuracy: 0.9883

It achieves the following results on the test set:
- Loss: 0.1359
- Precision: 0.92
- Recall: 0.9245
- F1: 0.9223
- Accuracy: 0.9785


### Framework versions

- Transformers 4.6.1
- Pytorch 1.8.1+cu101
- Datasets 1.6.2
- Tokenizers 0.10.2