initial commit

README.md

@@ -1,3 +1,37 @@
----
-license: cc-by-4.0
----
+---
+language:
+- da
+tags:
+- electra
+- pytorch
+- hatespeech
+license: cc-by-4.0
+datasets:
+- social media
+metrics:
+- f1
+widget:
+- text: "Senile gamle idiot"
+---
+
+# Danish ELECTRA for hate speech (offensive language) detection
+
+The ELECTRA Offensive model detects whether a Danish text is offensive or not. 
+It is based on the pretrained [Danish Ælæctra](Maltehb/aelaectra-danish-electra-small-cased) model. 
+
+See the [DaNLP documentation](https://danlp-alexandra.readthedocs.io/en/latest/docs/tasks/hatespeech.html#electra) for more details. 
+
+
+Here is how to use the model:
+
+```python
+from transformers import ElectraTokenizer, ElectraForSequenceClassification
+
+model = ElectraForSequenceClassification.from_pretrained("DaNLP/da-electra-hatespeech-detection")
+tokenizer = ElectraTokenizer.from_pretrained("DaNLP/da-electra-hatespeech-detection")
+```
+
+## Training data
+
+The data used for training has not been made publicly available. It consists of social media data manually annotated in collaboration with Danmarks Radio.
+

config.json

@@ -0,0 +1,37 @@
+{
+  "_name_or_path": ".",
+  "architectures": [
+    "ElectraForSequenceClassification"
+  ],
+  "id2label": {
+    "0": "not offensive",
+    "1": "offensive"
+  },
+  "label2id": {
+    "not offensive": 0,
+    "offensive": 1
+  },
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "embedding_size": 128,
+  "generator_size": "0.25",
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 256,
+  "initializer_range": 0.02,
+  "intermediate_size": 1024,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "electra",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "summary_activation": "gelu",
+  "summary_last_dropout": 0.1,
+  "summary_type": "first",
+  "summary_use_proj": true,
+  "transformers_version": "4.5.0",
+  "type_vocab_size": 2,
+  "vocab_size": 32000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62e42851587ac602a957534251bae9f3855468105a48c9f87218062286e9fdf2
+size 55043297

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, "special_tokens_map_file": null, "full_tokenizer_file": null, "model_max_length": 128, "name_or_path": "Maltehb/-l-ctra-danish-electra-small-cased", "do_basic_tokenize": true, "never_split": null}