File size: 3,512 Bytes
f01d363
 
 
c06e027
f01d363
 
 
 
9deda3f
f01d363
ad992d1
 
 
 
 
 
0bde925
ad992d1
0eb8992
c91902c
 
 
 
 
02e2481
c91902c
 
 
 
 
 
 
 
effac19
c91902c
 
 
 
 
 
1f50356
919b58f
 
0eb8992
15e720d
 
8dc5cc8
15e720d
ea7ff56
919b58f
f01d363
 
ad992d1
 
0bde925
ad992d1
4a03159
 
 
1f50356
919b58f
 
1f50356
effac19
4a03159
104babd
 
f01d363
104babd
 
 
 
1f50356
919b58f
1f50356
104babd
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
import plotly.express as px
import streamlit as st
from sentence_transformers import SentenceTransformer
from huggingface_hub import hf_hub_url, cached_download
import umap.umap_ as umap
import pandas as pd
import os
import joblib
import pkg_resources

def init_models():
    model_name = 'sentence-transformers/all-MiniLM-L6-v2'
    model = SentenceTransformer(model_name)
    REPO_ID = "peter2000/umap_embed_3d_all-MiniLM-L6-v2"
    FILENAME = "umap_embed_3d_all-MiniLM-L6-v2.sav"
    umap_model= joblib.load(cached_download(hf_hub_url(REPO_ID, FILENAME)))
    return model, umap_model

def app():

    with st.container():
        st.markdown("<h1 style='text-align: center;  \
                      color: black;'> Text Embedder</h1>", 
                      unsafe_allow_html=True)
       
        st.write(' ')
        st.write(' ')

    with st.expander("ℹ️ - About this app", expanded=True):

        st.write(
            """     
            Information cartography - Get your word/phrase/sentence/paragraph embedded and visualized.
            The (English) sentence-transformers model "all-MiniLM-L6-v2" maps sentences & paragraphs to a 384-dimensional dense vector space This is normally used for tasks like clustering or semantic search, but in this case, we use it to place your text to a 3D map. Before plotting, the dimension needs to be reduced to three so we can actually plot it, but preserve as much information as possible. For this, we use a technology called umap. The sentence transformer is context-sensitive and works best with whole sentences, to account for that we extend your text with "The book is about <text>".
            
            Simply put in your text and press EMBED, your examples will add up. You can use the category for different coloring.
            """)

        st.markdown("")
        
    word_to_embed_list =  st.session_state['embed_list']
    cat_list =  st.session_state['cat_list']
    
    with st.container():
        col1, col2 = st.columns(2)
        with col1:
            word_to_embed= st.text_input("Please enter your text here and we will embed it for you.", value="",)
        with col2:                                 
            cat= st.selectbox('Category',  ('1', '2', '3', '4', '5'))

        
        if st.button("Embed"):
            with st.spinner("👑 Embedding your input"):
                
                model, umap_model = init_models()
                
              
                word_to_embed_list.append(word_to_embed)
                
                st.session_state['embed_list'] = word_to_embed_list
                cat_list .append(cat)
                st.session_state['cat_list '] = cat_list 
                
                phrase_to_embed = ["The book is about "+ wte for wte in word_to_embed_list]
                examples_embeddings = model.encode(phrase_to_embed)
 
                examples_umap = umap_model.transform(examples_embeddings)

                #st.write(len(examples_umap))
                
                with st.spinner("👑 create visualisation"):  
                      fig = px.scatter_3d(
                          examples_umap[1:] , x=0, y=1, z=2,
                           color=cat_list[1:] ,
                          opacity = .7,    hover_data=[word_to_embed_list[1:]])
                      fig.update_scenes(xaxis_visible=False, yaxis_visible=False,zaxis_visible=False )
                      fig.update_traces(marker_size=4)
                      st.plotly_chart(fig)