Update README.md

README.md

@@ -4,88 +4,101 @@ tags:
 - sentence-transformers
 - feature-extraction
 - sentence-similarity
-
+license: mit
+datasets:
+- sentence-transformers/embedding-training-data
+- clips/mfaq
+- squad
+- eli5
+language:
+- da
+library_name: sentence-transformers
 ---
 
-# {MODEL_NAME}
+# MiniLM-L6-danish-encoder 
+
+This is a lightweight (~22 M parameters) [sentence-transformers](https://www.SBERT.net) model for Danish NLP: It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for tasks like clustering or semantic search. 
+
+The maximum sequence length is 128 tokens.
+
+The model was not pre-trained from scratch but adapted from the English version with a [tokenizer](https://huggingface.co/KennethTM/bert-base-uncased-danish) trained on Danish text.
+
+When using the model to retrieve relevant passages for a given query - "Query: " should be added to the query:
 
-This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 384 dimensional dense vector space and can be used for tasks like clustering or semantic search.
+```python
+query = "Kan man cykle på en vej?"
+query_template = f"Query: {query}"
 
-<!--- Describe your model here -->
+#query_template kan now be embedded and similarity compared to other passages
+```
 
-## Usage (Sentence-Transformers)
+# Usage (Sentence-Transformers)
 
 Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
 
 ```
 pip install -U sentence-transformers
 ```
-
 Then you can use the model like this:
 
 ```python
 from sentence_transformers import SentenceTransformer
-sentences = ["This is an example sentence", "Each sentence is converted"]
+sentences = ["Query: Kører der cykler på vejen?", "En mand løber på vejen.", "En panda løber på vejen.", "En mand kører hurtigt forbi på cykel."]
 
-model = SentenceTransformer('{MODEL_NAME}')
+model = SentenceTransformer('KennethTM/MiniLM-L6-danish-encoder')
 embeddings = model.encode(sentences)
 print(embeddings)
 ```
+# Usage (HuggingFace Transformers)
+Without [sentence-transformers](https://www.SBERT.net), you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.
 
+```python
+from transformers import AutoTokenizer, AutoModel
+import torch
+import torch.nn.functional as F
 
+#Mean Pooling - Take attention mask into account for correct averaging
+def mean_pooling(model_output, attention_mask):
+    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
+    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
 
-## Evaluation Results
-
-<!--- Describe how your model was evaluated -->
-
-For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
-
-
-## Training
-The model was trained with the parameters:
-
-**DataLoader**:
+# Sentences we want sentence embeddings for
+sentences = ["Query: Kører der cykler på vejen?", "En mand løber på vejen.", "En panda løber på vejen.", "En mand kører hurtigt forbi på cykel."]
 
-`torch.utils.data.dataloader.DataLoader` of length 14531 with parameters:
-```
-{'batch_size': 128, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}
-```
+# Load model from HuggingFace Hub
+tokenizer = AutoTokenizer.from_pretrained('KennethTM/MiniLM-L6-danish-encoder')
+model = AutoModel.from_pretrained('KennethTM/MiniLM-L6-danish-encoder')
 
-**Loss**:
+# Tokenize sentences
+encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
 
-`__main__.MultipleNegativesRankingLoss` with parameters:
-  ```
-  {'scale': 20.0, 'similarity_fct': 'cos_sim'}
-  ```
+# Compute token embeddings
+with torch.no_grad():
+    model_output = model(**encoded_input)
 
-Parameters of the fit()-Method:
-```
-{
-    "epochs": 1,
-    "evaluation_steps": 0,
-    "evaluator": "NoneType",
-    "max_grad_norm": 1,
-    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
-    "optimizer_params": {
-        "lr": 2e-05
-    },
-    "scheduler": "WarmupLinear",
-    "steps_per_epoch": null,
-    "warmup_steps": 1454,
-    "weight_decay": 0.01
-}
-```
+# Perform pooling
+sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
 
+# Normalize embeddings
+sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)
 
-## Full Model Architecture
+print("Sentence embeddings:")
+print(sentence_embeddings)
 ```
-SentenceTransformer(
-  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: BertModel 
-  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
-  (2): Normalize()
-)
-```
-
-## Citing & Authors
 
-<!--- Describe where people can find more information -->
+# Evaluation
+
+The performance of the pre-trained model was evaluated using [ScandEval](https://github.com/ScandEval/ScandEval).
+
+| Task                     | Dataset      | Score (±SE)                     |
+|:-------------------------|:-------------|:--------------------------------|
+| sentiment-classification | angry-tweets | mcc = 36.14 (±1.07)             |
+|                          |              | macro_f1 = 56.57 (±0.84)        |
+| named-entity-recognition | dane         | micro_f1 = 55.56 (±1.69)        |
+|                          |              | micro_f1_no_misc = 57.44 (±1.9) |
+| linguistic-acceptability | scala-da     | mcc = 12.4 (±3.07)              |
+|                          |              | macro_f1 = 53.54 (±2.15)        |
+| question-answering       | scandiqa-da  | em = 17.87 (±1.15)              |
+|                          |              | f1 = 27.84 (±1.37)              |
+| speed                    | speed        | speed = 18.59 (±0.05)           |