Add initial app version

.gitattributes

@@ -25,3 +25,11 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zstandard filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+documentstore_german-election-idx.pkl filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/config.json filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/config_sentence_transformers.json filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/modules.json filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/sentence_bert_config.json filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/special_tokens_map.json filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+adapted-retriever/tokenizer_config.json filter=lfs diff=lfs merge=lfs -text

adapted-retriever/.gitattributes

@@ -0,0 +1,2 @@
+pytorch_model.bin filter=lfs diff=lfs merge=lfs -text
+sentencepiece.bpe.model filter=lfs diff=lfs merge=lfs -text

adapted-retriever/1_Pooling/config.json

@@ -0,0 +1,7 @@
+{
+  "word_embedding_dimension": 768,
+  "pooling_mode_cls_token": false,
+  "pooling_mode_mean_tokens": true,
+  "pooling_mode_max_tokens": false,
+  "pooling_mode_mean_sqrt_len_tokens": false
+}

adapted-retriever/README.md

@@ -0,0 +1,125 @@
+---
+pipeline_tag: sentence-similarity
+tags:
+- sentence-transformers
+- feature-extraction
+- sentence-similarity
+- transformers
+---
+
+# {MODEL_NAME}
+
+This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
+
+<!--- Describe your model here -->
+
+## 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"]
+
+model = SentenceTransformer('{MODEL_NAME}')
+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
+
+
+#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)
+
+
+# Sentences we want sentence embeddings for
+sentences = ['This is an example sentence', 'Each sentence is converted']
+
+# Load model from HuggingFace Hub
+tokenizer = AutoTokenizer.from_pretrained('{MODEL_NAME}')
+model = AutoModel.from_pretrained('{MODEL_NAME}')
+
+# Tokenize sentences
+encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
+
+# Compute token embeddings
+with torch.no_grad():
+    model_output = model(**encoded_input)
+
+# Perform pooling. In this case, mean pooling.
+sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
+
+print("Sentence embeddings:")
+print(sentence_embeddings)
+```
+
+
+
+## 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**:
+
+`torch.utils.data.dataloader.DataLoader` of length 85 with parameters:
+```
+{'batch_size': 16, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}
+```
+
+**Loss**:
+
+`sentence_transformers.losses.MarginMSELoss.MarginMSELoss` 
+
+Parameters of the fit()-Method:
+```
+{
+    "epochs": 1,
+    "evaluation_steps": 0,
+    "evaluator": "NoneType",
+    "max_grad_norm": 1,
+    "optimizer_class": "<class 'transformers.optimization.AdamW'>",
+    "optimizer_params": {
+        "lr": 2e-05
+    },
+    "scheduler": "WarmupLinear",
+    "steps_per_epoch": null,
+    "warmup_steps": 8,
+    "weight_decay": 0.01
+}
+```
+
+
+## Full Model Architecture
+```
+SentenceTransformer(
+  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: XLMRobertaModel 
+  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
+)
+```
+
+## Citing & Authors
+
+<!--- Describe where people can find more information -->

adapted-retriever/config.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29703b29b31e2dabfcd73e52ba0856489249af29f2c8fc5209415fccadfac0d3
+size 821

adapted-retriever/config_sentence_transformers.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8c64b5cece00d8424b4896ea75b512b6008576088497609dfeb6bd63e6d36b8
+size 122

adapted-retriever/modules.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f4b264b80206c830bebbdcae377e137925650a433b689343a63bdc9b3145460
+size 229

adapted-retriever/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73203f5f04e88a22c1a336c4aceb89220dd8b1589151be83576767f781d2c00c
+size 1112244081

adapted-retriever/sentence_bert_config.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec8e29d6dcb61b611b7d3fdd2982c4524e6ad985959fa7194eacfb655a8d0d51
+size 53

adapted-retriever/sentencepiece.bpe.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
+size 5069051

adapted-retriever/special_tokens_map.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:378eb3bf733eb16e65792d7e3fda5b8a4631387ca04d2015199c4d4f22ae554d
+size 239

adapted-retriever/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46afe88da5fd71bdbab5cfab5e84c1adce59c246ea5f9341bbecef061891d0a7
+size 17082913

adapted-retriever/tokenizer_config.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c84cba673d65cd6fabcaf0340ae8e57b34306e01862132f4b476936917727dea
+size 483

app.py

@@ -0,0 +1,32 @@
+"""
+# My first app
+Here's our first attempt at using data to create a table:
+"""
+import streamlit as st
+import pandas as pd
+from load_css import local_css
+from retriever import do_search
+
+local_css('style.css')
+
+st.header('Semantic search demo')
+search = st.text_input('')
+
+if search:
+    result = do_search(search)
+    col1, col2, col3 = st.columns(3)
+
+    with col1:
+        st.write('TF-IDF')
+        st.write(result[0])
+
+    with col2:
+        st.write('Base dense retriever')
+        st.write(result[1])
+
+    with col3:
+        st.write('Adapted dense retriever')
+        st.write(result[2])
+
+
+

documentstore_german-election-idx.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc6e85c8a51b19f5b37df691bcfc75b57b1d24086b4e004489964a45927f9024
+size 4777552

requirements.txt

@@ -0,0 +1 @@
+farm-haystack

retriever.py

@@ -0,0 +1,69 @@
+from haystack.document_stores import InMemoryDocumentStore
+from haystack.utils import convert_files_to_docs
+from haystack.nodes.retriever import TfidfRetriever
+from haystack.pipelines import DocumentSearchPipeline, ExtractiveQAPipeline
+from haystack.nodes.retriever import EmbeddingRetriever
+from haystack.nodes import FARMReader
+import pickle
+from pprint import pprint
+
+class ExportableInMemoryDocumentStore(InMemoryDocumentStore):
+    """
+    Wrapper class around the InMemoryDocumentStore.
+    When the application is deployed to Huggingface Spaces there will be no GPU available.
+    We need to load pre-calculated data into the InMemoryDocumentStore.
+    """
+    def export(self, file_name='in_memory_store.pkl'):
+        with open(file_name, 'wb') as f:
+            pickle.dump(self.indexes, f)
+
+    def load_data(self, file_name='in_memory_store.pkl'):
+        with open(file_name, 'rb') as f:
+            self.indexes = pickle.load(f)
+
+
+
+document_store = ExportableInMemoryDocumentStore(similarity='cosine')
+document_store.load_data('documentstore_german-election-idx.pkl')
+
+retriever = TfidfRetriever(document_store=document_store)
+base_dense_retriever = EmbeddingRetriever(
+        document_store=document_store,
+        embedding_model='sentence-transformers/paraphrase-multilingual-mpnet-base-v2',
+        model_format='sentence_transformers'
+    )
+
+fine_tuned_retriever = EmbeddingRetriever(
+        document_store=document_store,
+        embedding_model='./adapted-retriever',
+        model_format='sentence_transformers'
+)
+
+def sparse_retrieval(query):
+    """Sparse retrieval pipeline"""
+    p_retrieval = DocumentSearchPipeline(retriever)
+    return p_retrieval.run(query=query)
+
+def dense_retrieval(query, retriever='base'):
+    if retriever == 'base':
+        p_retrieval = DocumentSearchPipeline(base_dense_retriever)
+    elif retriever == 'adapted':
+        p_retrieval = DocumentSearchPipeline(fine_tuned_retriever)
+    else:
+        return None
+
+    return p_retrieval.run(query=query)
+
+
+def do_search(query):
+    sparse_result = sparse_retrieval(query)['documents'][0].content
+    dense_base_result = dense_retrieval(query, retriever='base')['documents'][0].content
+    dense_adapted_result = dense_retrieval(query, retriever='adapted')['documents'][0].content
+    return sparse_result, dense_base_result, dense_adapted_result
+
+if __name__ == '__main__':
+    query = 'Klimawandel stoppen?'
+    result = do_search(query)
+    pprint(result)
+
+