Spaces:
Running
Running
anekameni
commited on
Commit
Β·
4a9aa3a
1
Parent(s):
e7ac6dd
Update README.md to enhance project overview, key features, and integration recommendations
Browse files
README.md
CHANGED
@@ -11,4 +11,55 @@ license: apache-2.0
|
|
11 |
short_description: Medivocate is an AI-driven platform leveraging Retrieval-Aug
|
12 |
---
|
13 |
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
short_description: Medivocate is an AI-driven platform leveraging Retrieval-Aug
|
12 |
---
|
13 |
|
14 |
+
# Medivocate
|
15 |
+
|
16 |
+
An AI-driven platform empowering users with trustworthy, personalized history guidance to combat misinformation and promote equitable history.
|
17 |
+
|
18 |
+
## Follows us [here](https://github.com/KameniAlexNea/medivocate)
|
19 |
+
|
20 |
+
* [**Alex Kameni**](https://www.linkedin.com/in/elie-alex-kameni-ngangue/)
|
21 |
+
* [**Esdras Fandio**](https://www.linkedin.com/in/esdras-fandio/)
|
22 |
+
* [**Patric Zeufack**](https://www.linkedin.com/in/zeufack-patric-hermann-7a9256143/)
|
23 |
+
|
24 |
+
## Project Overview
|
25 |
+
|
26 |
+
**Medivocate** is structured for modular development and ease of scalability, as seen in its directory layout:
|
27 |
+
|
28 |
+
```
|
29 |
+
π¦ ./
|
30 |
+
βββ π docs/
|
31 |
+
βββ π src/
|
32 |
+
β βββ π ocr/
|
33 |
+
β βββ π preprocessing/
|
34 |
+
β βββ π chunking/
|
35 |
+
β βββ π vector_store/
|
36 |
+
β βββ π rag_pipeline/
|
37 |
+
β βββ π llm_integration/
|
38 |
+
β βββ π prompt_engineering/
|
39 |
+
βββ π tests/
|
40 |
+
β βββ π unit/
|
41 |
+
β βββ π integration/
|
42 |
+
βββ π examples/
|
43 |
+
βββ π notebooks/
|
44 |
+
βββ π config/
|
45 |
+
βββ π README.md
|
46 |
+
βββ π CONTRIBUTING.md
|
47 |
+
βββ π requirements.txt
|
48 |
+
βββ π .gitignore
|
49 |
+
βββ π LICENSE
|
50 |
+
```
|
51 |
+
|
52 |
+
### Key Features
|
53 |
+
|
54 |
+
1. **Trustworthy Information Access** : Using RAG (Retrieval-Augmented Generation) pipelines to deliver fact-based responses.
|
55 |
+
2. **Advanced Document Handling** : Leveraging OCR, preprocessing, and chunking for scalable document ingestion.
|
56 |
+
3. **Integrated Tools** : Supports integration with vector databases (e.g., Chroma), LLMs, and advanced prompt engineering techniques.
|
57 |
+
|
58 |
+
### Recommendations for Integration
|
59 |
+
|
60 |
+
* **Groq** : Utilize Groq APIs for free-tier LLM support, perfect for prototyping RAG applications.
|
61 |
+
* **LangChain + LangSmith** : Build and monitor intelligent agents with LangChain and enhance debugging and evaluation using LangSmith.
|
62 |
+
* **Hugging Face Datasets** : For one-liner dataset loading and preprocessing, supporting efficient ML training pipelines.
|
63 |
+
* **Search Index** : Include Chroma for robust semantic search capabilities in RAG.
|
64 |
+
|
65 |
+
This modular design and extensive integration make Medivocate a powerful tool for historical education and research.
|