Upload 8 files

README.md

@@ -0,0 +1,73 @@
+---
+tags:
+- ner
+- token-classification
+- berturk
+- turkish
+language: tr
+datasets:
+- MilliyetNER
+widget:
+- text: "Türkiye'nin başkenti Ankara'dır ve ilk cumhurbaşkanı Mustafa Kemal Atatürk'tür."
+---
+
+# DATASET
+MilliyetNER dataset was collected from the Turkish Milliyet newspaper articles between 1997-1998. This dataset is presented by [Tür et al. (2003)](https://www.cambridge.org/core/journals/natural-language-engineering/article/abs/statistical-information-extraction-system-for-turkish/7C288FAFC71D5F0763C1F8CE66464017). It was collected from news articles and manually annotated with three different entity types: Person, Location, Organization. The authors did not provide training/validation/test splits for this dataset. Dataset splits used by [Yeniterzi et al. 2011](https://aclanthology.org/P11-3019).        
+For more information: [tdd.ai - MilliyetNER](https://data.tdd.ai/#/effafb5f-ebfc-4e5c-9a63-4f709ec1a135)    
+**Model is only trained using training set. Test set not included during the last training**.      
+
+# USAGE
+```python
+from transformers import pipeline, AutoModelForTokenClassification, AutoTokenizer
+model = AutoModelForTokenClassification.from_pretrained("alierenak/berturk-cased-ner")
+tokenizer = AutoTokenizer.from_pretrained("alierenak/berturk-cased-ner")
+ner_pipeline = pipeline('ner', model=model, tokenizer=tokenizer)
+ner_pipeline("Türkiye'nin başkenti Ankara, ilk cumhurbaşkanı Mustafa Kemal Atatürk'tür.")
+```
+
+# RESULT
+```bash
+[{'entity': 'B-LOCATION',
+  'score': 0.9966415,
+  'index': 1,
+  'word': 'Türkiye',
+  'start': 0,
+  'end': 7},
+ {'entity': 'B-LOCATION',
+  'score': 0.99456763,
+  'index': 5,
+  'word': 'Ankara',
+  'start': 21,
+  'end': 27},
+ {'entity': 'B-PERSON',
+  'score': 0.9958741,
+  'index': 9,
+  'word': 'Mustafa',
+  'start': 47,
+  'end': 54},
+ {'entity': 'I-PERSON',
+  'score': 0.98833394,
+  'index': 10,
+  'word': 'Kemal',
+  'start': 55,
+  'end': 60},
+ {'entity': 'I-PERSON',
+  'score': 0.9837286,
+  'index': 11,
+  'word': 'Atatürk',
+  'start': 61,
+  'end': 68}]
+```
+
+# BENCHMARKING
+```bash
+precision    recall  f1-score   support
+
+    LOCATION       0.97      0.96      0.97       960
+ORGANIZATION       0.95      0.92      0.94       863
+      PERSON       0.97      0.97      0.97      1410
+
+   micro avg       0.97      0.95      0.96      3233
+   macro avg       0.96      0.95      0.96      3233
+weighted avg       0.97      0.95      0.96      3233
+```

config.json

@@ -0,0 +1,43 @@
+{
+  "_name_or_path": "dbmdz/bert-base-turkish-cased",
+  "architectures": [
+    "BertForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "B-LOCATION",
+    "1": "B-ORGANIZATION",
+    "2": "B-PERSON",
+    "3": "I-LOCATION",
+    "4": "I-ORGANIZATION",
+    "5": "I-PERSON",
+    "6": "O"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "B-LOCATION": 0,
+    "B-ORGANIZATION": 1,
+    "B-PERSON": 2,
+    "I-LOCATION": 3,
+    "I-ORGANIZATION": 4,
+    "I-PERSON": 5,
+    "O": 6
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.9.2",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 32000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:488645a9d31fe8f4dd947a0003a1d9d7e01396a4052d83b00caf2d7e31caadcc
+size 134

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "max_len": 512, "special_tokens_map_file": null, "name_or_path": "dbmdz/bert-base-turkish-cased", "do_basic_tokenize": true, "never_split": null, "tokenizer_class": "BertTokenizer"}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:588c1a87288f783f6c29a6d0701f1b68a1a59baed8bd0586d836e14515955e22
+size 129