Upload README.md
Browse files
README.md
CHANGED
@@ -1,12 +1,54 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
https://huggingface.co/spaces/decodingdatascience/newrag-pine
|
3 |
+
|
4 |
+
# π Telecom Customer Support LLM with Groq API
|
5 |
+
|
6 |
+
This project demonstrates how to build a fast, production-grade AI-powered telecom customer support assistant using the **Groq API** and optimized GenAI configurations.
|
7 |
+
|
8 |
+
## π Project Overview
|
9 |
+
|
10 |
+
A step-by-step guide to:
|
11 |
+
- Sending POST requests using **Postman**
|
12 |
+
- Connecting to the **Groq API**
|
13 |
+
- Testing default vs. optimized GenAI configurations
|
14 |
+
- Applying structured **prompt templates**
|
15 |
+
- Deploying a simple LLM-powered support API
|
16 |
+
|
17 |
+
## π§ Tech Stack
|
18 |
+
|
19 |
+
| Layer | Tool/Tech |
|
20 |
+
|---------------|-------------------------|
|
21 |
+
| LLM | [Groq API](https://groq.com/) |
|
22 |
+
| API Platform | FastAPI / Postman |
|
23 |
+
| Prompt Design | Custom templates |
|
24 |
+
| Deployment | Localhost / Cloud (optional) |
|
25 |
+
|
26 |
+
## π§ AI Configuration
|
27 |
+
|
28 |
+
| Parameter | Description |
|
29 |
+
|---------------------|--------------------------------------|
|
30 |
+
| `temperature` | Controls randomness (default: 0.7) |
|
31 |
+
| `top_p` | Nucleus sampling |
|
32 |
+
| `max_tokens` | Max tokens to generate |
|
33 |
+
| `frequency_penalty` | Repetition control |
|
34 |
+
| `presence_penalty` | Topic diversity |
|
35 |
+
|
36 |
+
## π Experiment Setup
|
37 |
+
|
38 |
+
### 1. No Prompt Template + Default Config
|
39 |
+
- Basic user input
|
40 |
+
- Uses Groq defaults
|
41 |
+
- For benchmarking
|
42 |
+
|
43 |
+
### 2. With Prompt Template + Tuned Config
|
44 |
+
- Structured input (e.g., role, intent, constraints)
|
45 |
+
- Custom temperature and token limits
|
46 |
+
- Optimized for domain-specific responses
|
47 |
+
|
48 |
+
## π Quickstart
|
49 |
+
|
50 |
+
### Step 1: Clone the repo
|
51 |
+
|
52 |
+
```bash
|
53 |
+
git clone https://github.com/your-username/telecom-support-llm.git
|
54 |
+
cd telecom-support-llm
|