| Model Type | | language model, classification, question answering |
|
| Use Cases |
| Areas: | | Research, Commercial applications |
|
| Primary Use Cases: | | Text classification, Question Answering |
|
| Limitations: | | Limited to tasks in Bengali language |
|
|
| Additional Notes | | Similar performance to mBERT on some tasks |
|
| Supported Languages | |
| Training Details |
| Data Sources: | | https://oscar-corpus.com/, bn.wikipedia.org |
|
| Data Volume: | |
| Methodology: | | Pre-trained using ELECTRA methodology |
|
| Training Time: | | V1 - 120,000 steps; V2 - 190,000 steps |
|
| Model Architecture: | |
|