---
license: gpl-3.0
library_name: manas-1

# model-index:
#   - name: MANAS-1
#     results:
#       - task:
#           type: motor-imagery
#         dataset:
#           name: BCI-IV-2a
#           type: BCI-IV-2a
#         metrics:
#           - name: Balanced Accuracy
#             type: balanced_accuracy
#             value: 0.489
---

# MANAS-1
NeuroDX announces India's first Brain Language Model, MANAS-1. Designed as a Foundation Model for EEGs, MANAS-1 demonstrates SOTA performance across various tasks in BCI and clinical applications.
More details coming soon!

## Benchmarking Results:
More benchmark results will be added soon.

#### EEG-Bench Results
These metrics were computed using the EEG-Bench codebase. MANAS-1 and REVE (reve-base) were finetuned with a flatten-MLP setup. Finetuning setups for LaBraM and NeuroGPT were used as present in the EEG-Bench codebase.
| Dataset           | REVE               | MANAS-1             | LaBraM             | NeuroGPT          |
|-------------------|--------------------|--------------------|--------------------|-------------------|
| BCI-IV-2b         | <u>0.712 ± 0.01</u> | **0.723 ± 0.01**   | 0.707 ± 0.02       | 0.709             |
| Cho2017           | 0.526 ± 0.09       | <u>0.533 ± 0.02</u> | 0.504 ± 0.01       | **0.651 ± 0**     |
| BCI-IV-2a         | 0.434 ± 0.10       | **0.489 ± 0.03**   | 0.303 ± 0.04       | <u>0.461 ± 0.00</u> |
| Faller2012        | **0.566 ± 0.03**   | <u>0.560 ± 0.01</u> | 0.542 ± 0.01       | 0.513             |
| PhysioNet         | 0.348 ± 0.08       | <u>0.368 ± 0.01</u> | 0.284 ± 0.01       | **0.375 ± 0.00**  |
| Weibo2014         | **0.335 ± 0.04**   | <u>0.321 ± 0.01</u> | 0.254 ± 0.02       | 0.303             |
| Zhou2016          | 0.818 ± 0.16       | **0.865 ± 0.01**    | 0.687 ± 0.19       | <u>0.834</u>      |