karmiq commited on
Commit
083b8ca
1 Parent(s): 649bf8f

Update the README

Browse files
Files changed (1) hide show
  1. README.md +44 -33
README.md CHANGED
@@ -1,38 +1,43 @@
1
  ---
2
- language:
3
- - cs
4
- license:
5
- - cc-by-sa-3.0
6
- - gfdl
7
- size_categories:
8
- - 100K<n<1M
9
- task_categories:
10
- - text-generation
11
- - fill-mask
12
  dataset_info:
13
  features:
14
- - name: id
15
- dtype: string
16
- - name: url
17
- dtype: string
18
- - name: title
19
- dtype: string
20
- - name: chunks
21
- sequence: string
22
- - name: embeddings
23
- sequence:
24
- sequence: float32
25
  splits:
26
- - name: train
27
- num_bytes: 2580729273
28
- num_examples: 534044
29
- download_size: 2307157994
30
  dataset_size: 2580729273
 
31
  configs:
32
- - config_name: default
33
- data_files:
34
- - split: train
35
- path: data/train-*
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36
  ---
37
 
38
  This dataset contains the Czech subset of the [`wikimedia/wikipedia`](https://huggingface.co/datasets/wikimedia/wikipedia) dataset. Each page is divided into paragraphs, stored as a list in the `chunks` column. For every paragraph, embeddings are created using the [`Seznam/simcse-dist-mpnet-paracrawl-cs-en`](https://huggingface.co/Seznam/simcse-dist-mpnet-paracrawl-cs-en) model.
@@ -107,8 +112,14 @@ import os
107
  import textwrap
108
 
109
  import sentence_transformers
 
 
 
 
110
 
111
- model = sentence_transformers.SentenceTransformer("Seznam/simcse-dist-mpnet-paracrawl-cs-en")
 
 
112
 
113
  ds.set_format(type="torch", columns=["embeddings"], output_all_columns=True)
114
 
@@ -147,13 +158,13 @@ for hit in hits[0]:
147
  chunk = ds_flat[hit['corpus_id']]['chunk']
148
  print(f"[{hit['score']:0.2f}] {textwrap.shorten(chunk, width=100, placeholder='…')} [{title}]")
149
 
150
- # [0.58] Dynamika Fyzikální zákony [Newtonovy pohybové zákony]
151
- # [0.53] Teorie množin [Ordinální číslo]
152
  # ...
153
  ```
154
  </details>
155
 
156
- The embeddings generation took about 25 minutes on an NVIDIA RTX 4090 24GB.
157
 
158
  ## License
159
 
 
1
  ---
 
 
 
 
 
 
 
 
 
 
2
  dataset_info:
3
  features:
4
+ - name: id
5
+ dtype: string
6
+ - name: url
7
+ dtype: string
8
+ - name: title
9
+ dtype: string
10
+ - name: chunks
11
+ sequence: string
12
+ - name: embeddings
13
+ sequence:
14
+ sequence: float32
15
  splits:
16
+ - name: train
17
+ num_bytes: 2580729273
18
+ num_examples: 534044
19
+ download_size: 2307703671
20
  dataset_size: 2580729273
21
+
22
  configs:
23
+ - config_name: default
24
+ data_files:
25
+ - split: train
26
+ path: data/train-*
27
+
28
+ language:
29
+ - cs
30
+
31
+ size_categories:
32
+ - 100K<n<1M
33
+
34
+ task_categories:
35
+ - text-generation
36
+ - fill-mask
37
+
38
+ license:
39
+ - cc-by-sa-3.0
40
+ - gfdl
41
  ---
42
 
43
  This dataset contains the Czech subset of the [`wikimedia/wikipedia`](https://huggingface.co/datasets/wikimedia/wikipedia) dataset. Each page is divided into paragraphs, stored as a list in the `chunks` column. For every paragraph, embeddings are created using the [`Seznam/simcse-dist-mpnet-paracrawl-cs-en`](https://huggingface.co/Seznam/simcse-dist-mpnet-paracrawl-cs-en) model.
 
112
  import textwrap
113
 
114
  import sentence_transformers
115
+ from sentence_transformers.models import Transformer, Pooling
116
+
117
+ from sentence_transformers import SentenceTransformer
118
+ from sentence_transformers.models import Transformer, Pooling
119
 
120
+ embedding_model = Transformer("Seznam/simcse-dist-mpnet-paracrawl-cs-en")
121
+ pooling = Pooling(word_embedding_dimension=embedding_model.get_word_embedding_dimension(), pooling_mode="cls")
122
+ model = SentenceTransformer(modules=[embedding_model, pooling])
123
 
124
  ds.set_format(type="torch", columns=["embeddings"], output_all_columns=True)
125
 
 
158
  chunk = ds_flat[hit['corpus_id']]['chunk']
159
  print(f"[{hit['score']:0.2f}] {textwrap.shorten(chunk, width=100, placeholder='…')} [{title}]")
160
 
161
+ # [0.72] Molekulová fyzika ( též molekulární fyzika ) je část fyziky, která zkoumá látky na úrovni atomů a… [Molekulová fyzika]
162
+ # [0.70] Fyzika ( z řeckého φυσικός ( fysikos ): přírodní, ze základu φύσις ( fysis ): příroda, archaicky… [Fyzika]
163
  # ...
164
  ```
165
  </details>
166
 
167
+ The embeddings generation took about 35 minutes on an NVIDIA A100 80GB.
168
 
169
  ## License
170