lilbablo commited on
Commit
50ecb57
·
verified ·
1 Parent(s): 54d45e9

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. README.md +72 -0
  2. requirements.txt +5 -0
README.md ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # 🧠 Financial Literacy Quiz — Smart Money Assistant
2
+
3
+ This project is an AI-powered financial literacy quiz designed to help users build better money habits through realistic scenarios and immediate feedback.
4
+
5
+ The assistant, named **Smart Money**, is powered by a fine-tuned [Phi-2](https://huggingface.co/microsoft/phi-2) model using LoRA (Low-Rank Adaptation). It evaluates your financial decisions and gives witty, sharp explanations when your choices fall short.
6
+
7
+ ---
8
+
9
+ ## 📦 Included
10
+
11
+ - `cli_quiz.py` — Command-line script to run the quiz
12
+ - `quiz_ready_dataset.jsonl` — A custom dataset of 1000+ realistic financial scenarios
13
+ - `model/` — Fine-tuned Phi-2 model directory (LoRA-merged or base)
14
+ - `README.md` — Project documentation
15
+
16
+ ---
17
+
18
+ ## 🚀 Quick Start
19
+
20
+ To run the quiz locally:
21
+
22
+ \`\`\`bash
23
+ pip install torch transformers
24
+ python3 cli_quiz.py
25
+ \`\`\`
26
+
27
+ > Note: If you are not using a GPU, modify the script to use `torch.float32` instead of `torch.float16`.
28
+
29
+ ---
30
+
31
+ ## 💡 How It Works
32
+
33
+ 1. You’ll be asked 5 randomly selected financial scenario questions.
34
+ 2. Choose your best response (multiple choice: a, b, or c).
35
+ 3. If you’re wrong, Smart Money replies with a witty one-liner.
36
+ 4. At the end, you receive a final score and personalized summary.
37
+
38
+ ---
39
+
40
+ ## 🧠 Model Details
41
+
42
+ - **Base Model**: [Phi-2 by Microsoft](https://huggingface.co/microsoft/phi-2)
43
+ - **Fine-Tuning Method**: LoRA via `peft`
44
+ - **Libraries Used**: `transformers`, `torch`, `json`, `random`
45
+ - **Inference**: Local GPU (tested on 2x RTX 4080)
46
+
47
+ ---
48
+
49
+ ## 🤖 Why This Matters
50
+
51
+ Many people lack access to practical, relatable financial education. This project bridges that gap by:
52
+
53
+ - Using an LLM to simulate a sharp financial coach
54
+ - Teaching through interactive, contextual scenarios
55
+ - Reinforcing good financial habits with real-time feedback
56
+
57
+ ---
58
+
59
+ ## ✍️ Created By
60
+
61
+ [@lilbablo](https://huggingface.co/lilbablo) — exploring the intersection of AI, financial literacy, and interactive learning.
62
+
63
+ ---
64
+
65
+ ## 🤝 Contribute
66
+
67
+ Pull requests welcome! You can:
68
+
69
+ - Add more scenario questions
70
+ - Improve model responses
71
+ - Port this to Gradio or Streamlit
72
+ - Translate the quiz for international use
requirements.txt ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ torch>=2.0.0
2
+ transformers>=4.36.0
3
+ peft>=0.8.0
4
+ accelerate>=0.24.0
5
+ bitsandbytes>=0.41.0