Fill-Mask
Transformers
Safetensors
nandi_encoder
encoder
masked-language-modeling
indic
multilingual
code
efficient
frontiersmind
nandi
custom_code
Instructions to use FrontiersMind/NandiBERT-150M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use FrontiersMind/NandiBERT-150M with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="FrontiersMind/NandiBERT-150M", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForMaskedLM model = AutoModelForMaskedLM.from_pretrained("FrontiersMind/NandiBERT-150M", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Model cards: use full names with size for compared models
Browse files
README.md
CHANGED
|
@@ -74,7 +74,7 @@ NandiBERT is a family of open bidirectional encoders that reach state of the art
|
|
| 74 |
<tr><th>IndicSentiment</th><th>IndicXNLI</th><th>IndicCOPA</th><th>IndicXParaphrase</th><th>Naamapadam NER (f1)</th><th>MASSIVE Slot-fill (f1)</th><th>IndicQA (f1)</th></tr>
|
| 75 |
</thead>
|
| 76 |
<tbody>
|
| 77 |
-
<tr><td>
|
| 78 |
<tr><td>NandiBERT-150M</td><td>149M</td><td>91.3</td><td>75.9</td><td>66.4</td><td>60.6</td><td>77.4</td><td>54</td><td>46</td></tr>
|
| 79 |
</tbody>
|
| 80 |
</table>
|
|
@@ -87,14 +87,14 @@ NandiBERT is a family of open bidirectional encoders that reach state of the art
|
|
| 87 |
<tr><th>GLUE</th><th>SQuAD 2.0</th><th>MUltiCoNERv2</th><th>few-nerd</th><th>Reward Bench</th><th>BEIR</th><th>RAR-b Code</th><th>RAR-b Math</th><th>MLDR</th><th>LongEmbed</th></tr>
|
| 88 |
</thead>
|
| 89 |
<tbody>
|
| 90 |
-
<tr><td>
|
| 91 |
-
<tr><td>
|
| 92 |
-
<tr><td>MosaicBERT</td><td>English</td><td>137M</td><td>85.4</td><td>79.4</td><td>57.8</td><td>65.7</td><td>62.5</td><td>41.4</td><td>12.7</td><td>30.9</td><td>29.6</td><td>40.5</td><td>54.48</td></tr>
|
| 93 |
-
<tr><td>NomicBERT</td><td>English</td><td>137M</td><td>84</td><td>82.7</td><td>60.1</td><td>66.9</td><td>64.1</td><td>42</td><td>13.8</td><td>30.2</td><td>29.9</td><td>45</td><td>55.48</td></tr>
|
| 94 |
-
<tr><td>GTE-en-MLM</td><td>English</td><td>137M</td><td>85.6</td><td>82.7</td><td>59.2</td><td>66.4</td><td>63.4</td><td>42.4</td><td>12</td><td>33.6</td><td>37.1</td><td>58.5</td><td>55.66</td></tr>
|
| 95 |
-
<tr><td>
|
| 96 |
-
<tr><td>Ettin
|
| 97 |
-
<tr><td>UltraBERT</td><td>English</td><td>152M</td><td>89.5</td><td>89.2</td><td>57.9</td><td>67.4</td><td>73</td><td>45.2</td><td>39.1</td><td>55.4</td><td>43.5</td><td>72</td><td>64.59</td></tr>
|
| 98 |
<tr><td>NandiBERT-150M</td><td>Multi-Lingual</td><td>149M</td><td>87.1</td><td>87.6</td><td>56.9</td><td>68.1</td><td>71.2</td><td>44.5</td><td>32.8</td><td>49.4</td><td>41.2</td><td>71.1</td><td>62.20</td></tr>
|
| 99 |
</tbody>
|
| 100 |
</table>
|
|
|
|
| 74 |
<tr><th>IndicSentiment</th><th>IndicXNLI</th><th>IndicCOPA</th><th>IndicXParaphrase</th><th>Naamapadam NER (f1)</th><th>MASSIVE Slot-fill (f1)</th><th>IndicQA (f1)</th></tr>
|
| 75 |
</thead>
|
| 76 |
<tbody>
|
| 77 |
+
<tr><td>IndicBERT-v2</td><td>272M</td><td>91.2</td><td>73.7</td><td>64.6</td><td>56.9</td><td>73.2</td><td>58.1</td><td>48.3</td></tr>
|
| 78 |
<tr><td>NandiBERT-150M</td><td>149M</td><td>91.3</td><td>75.9</td><td>66.4</td><td>60.6</td><td>77.4</td><td>54</td><td>46</td></tr>
|
| 79 |
</tbody>
|
| 80 |
</table>
|
|
|
|
| 87 |
<tr><th>GLUE</th><th>SQuAD 2.0</th><th>MUltiCoNERv2</th><th>few-nerd</th><th>Reward Bench</th><th>BEIR</th><th>RAR-b Code</th><th>RAR-b Math</th><th>MLDR</th><th>LongEmbed</th></tr>
|
| 88 |
</thead>
|
| 89 |
<tbody>
|
| 90 |
+
<tr><td>BERT-base</td><td>English</td><td>110M</td><td>84.7</td><td>76.3</td><td>53.8</td><td>66.6</td><td>59.4</td><td>36.7</td><td>6.6</td><td>27.5</td><td>29</td><td>29.3</td><td>51.45</td></tr>
|
| 91 |
+
<tr><td>RoBERTa-base</td><td>English</td><td>125M</td><td>86.4</td><td>83.7</td><td>56.5</td><td>67.1</td><td>63.2</td><td>39.6</td><td>11.7</td><td>33.4</td><td>28.7</td><td>33.8</td><td>55.20</td></tr>
|
| 92 |
+
<tr><td>MosaicBERT-base</td><td>English</td><td>137M</td><td>85.4</td><td>79.4</td><td>57.8</td><td>65.7</td><td>62.5</td><td>41.4</td><td>12.7</td><td>30.9</td><td>29.6</td><td>40.5</td><td>54.48</td></tr>
|
| 93 |
+
<tr><td>NomicBERT-2048</td><td>English</td><td>137M</td><td>84</td><td>82.7</td><td>60.1</td><td>66.9</td><td>64.1</td><td>42</td><td>13.8</td><td>30.2</td><td>29.9</td><td>45</td><td>55.48</td></tr>
|
| 94 |
+
<tr><td>GTE-en-MLM-base</td><td>English</td><td>137M</td><td>85.6</td><td>82.7</td><td>59.2</td><td>66.4</td><td>63.4</td><td>42.4</td><td>12</td><td>33.6</td><td>37.1</td><td>58.5</td><td>55.66</td></tr>
|
| 95 |
+
<tr><td>ModernBERT-base</td><td>English</td><td>149M</td><td>88.4</td><td>86.7</td><td>57.4</td><td>66.7</td><td>68</td><td>42.6</td><td>29.6</td><td>46.1</td><td>37.7</td><td>63.2</td><td>60.69</td></tr>
|
| 96 |
+
<tr><td>Ettin-Encoder-150M</td><td>English</td><td>149M</td><td>88.9</td><td>87.4</td><td>57.4</td><td>66.5</td><td>70.3</td><td>42.4</td><td>31.8</td><td>47.1</td><td>35.9</td><td>60.1</td><td>61.48</td></tr>
|
| 97 |
+
<tr><td>UltraBERT-150M</td><td>English</td><td>152M</td><td>89.5</td><td>89.2</td><td>57.9</td><td>67.4</td><td>73</td><td>45.2</td><td>39.1</td><td>55.4</td><td>43.5</td><td>72</td><td>64.59</td></tr>
|
| 98 |
<tr><td>NandiBERT-150M</td><td>Multi-Lingual</td><td>149M</td><td>87.1</td><td>87.6</td><td>56.9</td><td>68.1</td><td>71.2</td><td>44.5</td><td>32.8</td><td>49.4</td><td>41.2</td><td>71.1</td><td>62.20</td></tr>
|
| 99 |
</tbody>
|
| 100 |
</table>
|