shantipriya
commited on
Commit
•
8bc7f73
1
Parent(s):
56d3381
Update README.md
Browse files
README.md
CHANGED
@@ -7,4 +7,51 @@ sdk: static
|
|
7 |
pinned: false
|
8 |
---
|
9 |
|
10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
pinned: false
|
8 |
---
|
9 |
|
10 |
+
## About
|
11 |
+
The Odia Generative AI (in short, OdiaGenAI) is an initiative to research Generative AI and Large Language Models (LLMs) for the low-resource Odia language.
|
12 |
+
|
13 |
+
## Objective
|
14 |
+
The OdiaGenAI aims to
|
15 |
+
1. Build pre-trained Odia LLM,
|
16 |
+
2. Fine-tuned Odia LLM, and
|
17 |
+
3. Instruct LLM (Odia).
|
18 |
+
|
19 |
+
The data, code, and models will be available to the public for research and non-commercial purposes.
|
20 |
+
|
21 |
+
## Why OdiaGenAI
|
22 |
+
|
23 |
+
* **First**: Though many LLMs support multilingual, including Odia language, the performance for various tasks (e.g., content generation, question-answering) is limited due to the amount of ingested data for Odia.
|
24 |
+
* **Second**: There is subscription or fees associated with the high-performing LLMs.
|
25 |
+
|
26 |
+
* **Third**: The usage (privacy) and bias of data input to these LLMs are in question.
|
27 |
+
|
28 |
+
## What are the focus research areas of OdiaGenAI
|
29 |
+
We have divided the primary focus areas into three parts.
|
30 |
+
|
31 |
+
**1. Literature Survey:** Investigate the latest developments in Generative AI and LLMs and analyze current methods to support the Odia language for different tasks.
|
32 |
+
|
33 |
+
**2. Development:** Developing pre-trained and fine-tuned Odia LLM, which includes dataset preparation, model training, evaluation, prompt engineering, and API development.
|
34 |
+
|
35 |
+
**3. Deployment:** Deploy the Odia LLM models for public access for research and non-commercial purposes.
|
36 |
+
|
37 |
+
## Who can use OdiaGenAI LLMs
|
38 |
+
The models (pre-trained/fine-tuned) will be available through Hugging Face for research and non-commercial purposes. Feel free to contact us for a domain-specific application or particular use cases.
|
39 |
+
|
40 |
+
## What are the use cases of OdiaGenAI LLMs
|
41 |
+
There are several use cases of OdiaGenAI LLMs. Three primary domains relating to Odisha which we are focusing to use the developed LLM are:
|
42 |
+
* Education
|
43 |
+
* Healthcare
|
44 |
+
* Governance
|
45 |
+
* Tourism
|
46 |
+
* Agriculture
|
47 |
+
* Industrial Application
|
48 |
+
|
49 |
+
## Contributors
|
50 |
+
* [Shantipriya Parida](https://www.linkedin.com/in/shantipriya-parida-9781a9127/)
|
51 |
+
* [Sambit Sekhar](https://www.linkedin.com/in/sambit-sekhar-ai/)
|
52 |
+
* [Subhadarshi Panda](https://www.linkedin.com/in/subhadarshi-panda-1ba5091a/)
|
53 |
+
* [Soumendra Kumar Sahoo](https://www.linkedin.com/in/soumendrak/)
|
54 |
+
* [Swateek Jena](https://www.linkedin.com/in/swateek/)
|
55 |
+
* [Abhijeet Parida](https://www.linkedin.com/in/a-parida/)
|
56 |
+
* [Arghyadeep Sen](https://www.linkedin.com/in/arghyadeep-sen-kiit/)
|
57 |
+
* [Dr. Satya Ranjan Dash](https://ksca.kiit.ac.in/profiles/satya-ranjan-dash/)
|