Update README.md
Browse files
README.md
CHANGED
|
@@ -13,3 +13,198 @@ language:
|
|
| 13 |
base_model:
|
| 14 |
- FacebookAI/xlm-roberta-large
|
| 15 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
base_model:
|
| 14 |
- FacebookAI/xlm-roberta-large
|
| 15 |
---
|
| 16 |
+
|
| 17 |
+
<table style="width:100%;height:100%">
|
| 18 |
+
<tr>
|
| 19 |
+
<td colspan=2>
|
| 20 |
+
<h4><i class="highlight-container"><b class="highlight">TeslaXLM</b></i></h4>
|
| 21 |
+
</td>
|
| 22 |
+
</tr>
|
| 23 |
+
<tr style="width:100%;height:100%">
|
| 24 |
+
<td width=50%>
|
| 25 |
+
<p>Вишејезични модел, 561 милион параметара</p>
|
| 26 |
+
<p>Обучаван над корпусима српског и српскохрватског језика - 20 милијарди речи</p>
|
| 27 |
+
<p>Једнака подршка уноса на ћирилици и латиници!</p>
|
| 28 |
+
</td>
|
| 29 |
+
<td>
|
| 30 |
+
<p>Multilingual model, 561 million parameters</p>
|
| 31 |
+
<p>Trained on Serbian and Serbo-Croatian corpora - 20 billion words</p>
|
| 32 |
+
<p>Equal support for Cyrillic and Latin input!</p>
|
| 33 |
+
</td>
|
| 34 |
+
</tr>
|
| 35 |
+
</table>
|
| 36 |
+
|
| 37 |
+
```python
|
| 38 |
+
>>> from transformers import pipeline
|
| 39 |
+
>>> unmasker = pipeline('fill-mask', model='te-sla/teslaXLM')
|
| 40 |
+
>>> unmasker("Kada bi čovek znao gde će pasti on bi<mask>.")
|
| 41 |
+
```
|
| 42 |
+
|
| 43 |
+
```python
|
| 44 |
+
>>> from transformers import AutoTokenizer, AutoModelForMaskedLM
|
| 45 |
+
>>> from torch import LongTensor, no_grad
|
| 46 |
+
>>> from scipy import spatial
|
| 47 |
+
>>> tokenizer = AutoTokenizer.from_pretrained('te-sla/teslaXLM')
|
| 48 |
+
>>> model = AutoModelForMaskedLM.from_pretrained('te-sla/teslaXLM', output_hidden_states=True)
|
| 49 |
+
>>> x = " pas"
|
| 50 |
+
>>> y = " mačka"
|
| 51 |
+
>>> z = " svemir"
|
| 52 |
+
>>> tensor_x = LongTensor(tokenizer.encode(x, add_special_tokens=False)).unsqueeze(0)
|
| 53 |
+
>>> tensor_y = LongTensor(tokenizer.encode(y, add_special_tokens=False)).unsqueeze(0)
|
| 54 |
+
>>> tensor_z = LongTensor(tokenizer.encode(z, add_special_tokens=False)).unsqueeze(0)
|
| 55 |
+
>>> model.eval()
|
| 56 |
+
>>> with no_grad():
|
| 57 |
+
>>> vektor_x = model(input_ids=tensor_x).hidden_states[-1].squeeze()
|
| 58 |
+
>>> vektor_y = model(input_ids=tensor_y).hidden_states[-1].squeeze()
|
| 59 |
+
>>> vektor_z = model(input_ids=tensor_z).hidden_states[-1].squeeze()
|
| 60 |
+
>>> print(spatial.distance.cosine(vektor_x, vektor_y))
|
| 61 |
+
>>> print(spatial.distance.cosine(vektor_x, vektor_z))
|
| 62 |
+
```
|
| 63 |
+
|
| 64 |
+
<div class="inline-flex flex-col" style="line-height: 1.5;padding-right:50px">
|
| 65 |
+
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">Author</div>
|
| 66 |
+
<a href="https://huggingface.co/procesaur">
|
| 67 |
+
<div class="flex">
|
| 68 |
+
<div
|
| 69 |
+
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%;
|
| 70 |
+
background-size: cover; background-image: url('https://cdn-uploads.huggingface.co/production/uploads/1673534533167-63bc254fb8c61b8aa496a39b.jpeg?w=200&h=200&f=face')">
|
| 71 |
+
</div>
|
| 72 |
+
</div>
|
| 73 |
+
</a>
|
| 74 |
+
<div style="text-align: center; font-size: 16px; font-weight: 800">Mihailo Škorić</div>
|
| 75 |
+
<div>
|
| 76 |
+
<a href="https://huggingface.co/procesaur">
|
| 77 |
+
<div style="text-align: center; font-size: 14px;">@procesaur</div>
|
| 78 |
+
</a>
|
| 79 |
+
</div>
|
| 80 |
+
</div>
|
| 81 |
+
</div>
|
| 82 |
+
|
| 83 |
+
<div class="inline-flex flex-col" style="line-height: 1.5;padding-right:50px">
|
| 84 |
+
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">Author</div>
|
| 85 |
+
<a href="https://huggingface.co/tanor">
|
| 86 |
+
<div class="flex">
|
| 87 |
+
<div
|
| 88 |
+
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%;
|
| 89 |
+
background-size: cover; background-image: url('https://cdn-avatars.huggingface.co/v1/production/uploads/6409d3d71ee054d66a673701/KTOOnCRS9NhpAMZIvLlU7.png?w=200&h=200&f=face')">
|
| 90 |
+
</div>
|
| 91 |
+
</div>
|
| 92 |
+
</a>
|
| 93 |
+
<div style="text-align: center; font-size: 16px; font-weight: 800">Saša Petalinkar</div>
|
| 94 |
+
<div>
|
| 95 |
+
<a href="https://huggingface.co/tanor">
|
| 96 |
+
<div style="text-align: center; font-size: 14px;">@tanor</div>
|
| 97 |
+
</a>
|
| 98 |
+
</div>
|
| 99 |
+
</div>
|
| 100 |
+
</div>
|
| 101 |
+
|
| 102 |
+
<div class="inline-flex flex-col" style="line-height: 1.5;">
|
| 103 |
+
<div style="text-align: center; margin-top: 3px; font-size: 16px; font-weight: 800">Computation</div>
|
| 104 |
+
<a href="https://tesla.rgf.bg.ac.rs">
|
| 105 |
+
<div class="flex">
|
| 106 |
+
<div
|
| 107 |
+
style="display:DISPLAY_1; margin-left: auto; margin-right: auto; width: 92px; height:92px; border-radius: 50%;
|
| 108 |
+
background-size: cover; background-image: url(https://cdn-avatars.huggingface.co/v1/production/uploads/63bc254fb8c61b8aa496a39b/TfM_-sc8-b34ddfhHBGTA.png?w=200&h=200&f=face)">
|
| 109 |
+
</div>
|
| 110 |
+
</div>
|
| 111 |
+
</a>
|
| 112 |
+
<div style="text-align: center; font-size: 16px; font-weight: 800">TESLA project</div>
|
| 113 |
+
<div>
|
| 114 |
+
<a href="https://huggingface.co/te-sla">
|
| 115 |
+
<div style="text-align: center; font-size: 14px;">@te-sla</div>
|
| 116 |
+
</a>
|
| 117 |
+
</div>
|
| 118 |
+
</div>
|
| 119 |
+
</div>
|
| 120 |
+
<br/><br/>
|
| 121 |
+
<div id="zastava">
|
| 122 |
+
<div class="grb">
|
| 123 |
+
<img src="https://www.ai.gov.rs/img/logo_60x120-2.png" style="position:relative; left:30px; z-index:10; height:85px">
|
| 124 |
+
</div>
|
| 125 |
+
<table width=100% style="border:0px">
|
| 126 |
+
<tr style="background-color:#C6363C;width:100%;border:0px;height:30px"><td style="width:100vw"></td></tr>
|
| 127 |
+
<tr style="background-color:#0C4076;width:100%;border:0px;height:30px"><td></td></tr>
|
| 128 |
+
<tr style="background-color:#ffffff;width:100%;border:0px;height:30px"><td></td></tr>
|
| 129 |
+
</table>
|
| 130 |
+
</div>
|
| 131 |
+
|
| 132 |
+
<table style="width:100%;height:100%">
|
| 133 |
+
<tr style="width:100%;height:100%">
|
| 134 |
+
<td width=50%>
|
| 135 |
+
<p>Истраживање jе спроведено уз подршку Фонда за науку Републике Србиjе, #7276, Text Embeddings – Serbian Language Applications – TESLA</p>
|
| 136 |
+
</td>
|
| 137 |
+
<td>
|
| 138 |
+
<p>This research was supported by the Science Fund of the Republic of Serbia, #7276, Text Embeddings - Serbian Language Applications - TESLA</p>
|
| 139 |
+
</td>
|
| 140 |
+
</tr>
|
| 141 |
+
</table>
|
| 142 |
+
|
| 143 |
+
|
| 144 |
+
|
| 145 |
+
<style>
|
| 146 |
+
.ffeat: {
|
| 147 |
+
color:red
|
| 148 |
+
}
|
| 149 |
+
|
| 150 |
+
.cover {
|
| 151 |
+
width: 100%;
|
| 152 |
+
margin-bottom: 5pt
|
| 153 |
+
}
|
| 154 |
+
|
| 155 |
+
.highlight-container, .highlight {
|
| 156 |
+
position: relative;
|
| 157 |
+
text-decoration:none
|
| 158 |
+
}
|
| 159 |
+
|
| 160 |
+
.highlight-container {
|
| 161 |
+
display: inline-block;
|
| 162 |
+
|
| 163 |
+
}
|
| 164 |
+
|
| 165 |
+
.highlight{
|
| 166 |
+
color:white;
|
| 167 |
+
text-transform:uppercase;
|
| 168 |
+
font-size: 16pt;
|
| 169 |
+
}
|
| 170 |
+
|
| 171 |
+
.highlight-container{
|
| 172 |
+
padding:5px 10px
|
| 173 |
+
}
|
| 174 |
+
|
| 175 |
+
.highlight-container:before {
|
| 176 |
+
content: " ";
|
| 177 |
+
display: block;
|
| 178 |
+
height: 100%;
|
| 179 |
+
width: 100%;
|
| 180 |
+
margin-left: 0px;
|
| 181 |
+
margin-right: 0px;
|
| 182 |
+
position: absolute;
|
| 183 |
+
background: #e80909;
|
| 184 |
+
transform: rotate(2deg);
|
| 185 |
+
top: -1px;
|
| 186 |
+
left: -1px;
|
| 187 |
+
border-radius: 20% 25% 20% 24%;
|
| 188 |
+
padding: 10px 18px 18px 10px;
|
| 189 |
+
}
|
| 190 |
+
|
| 191 |
+
div.grb, #zastava>table {
|
| 192 |
+
position:absolute;
|
| 193 |
+
top:0px;
|
| 194 |
+
left: 0px;
|
| 195 |
+
margin:0px
|
| 196 |
+
}
|
| 197 |
+
|
| 198 |
+
div.grb>img, #zastava>table{
|
| 199 |
+
margin:0px
|
| 200 |
+
}
|
| 201 |
+
|
| 202 |
+
#zastava {
|
| 203 |
+
position: relative;
|
| 204 |
+
margin-bottom:120px
|
| 205 |
+
}
|
| 206 |
+
|
| 207 |
+
p {
|
| 208 |
+
font-size:14pt
|
| 209 |
+
}
|
| 210 |
+
</style>
|