Update README.md
Browse files
README.md
CHANGED
|
@@ -23,23 +23,38 @@ The model demonstrated stable and consistent improvement throughout the training
|
|
| 23 |
|
| 24 |
The true power of the fine-tuned model is its ability to understand semantic context beyond simple keyword matching. In the following challenging example, the fine-tuned model correctly infers the answer, while the original base model fails.
|
| 25 |
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
**Analysis**: The fine-tuned model correctly identifies 'Tesla' by understanding the semantic relationship between the query and the document, even with no direct keyword overlap. In contrast, the original model is easily confused by distractors and fails to rank the correct answer first, demonstrating the significant impact of the ColBERT fine-tuning process.
|
| 45 |
|
|
|
|
| 23 |
|
| 24 |
The true power of the fine-tuned model is its ability to understand semantic context beyond simple keyword matching. In the following challenging example, the fine-tuned model correctly infers the answer, while the original base model fails.
|
| 25 |
|
| 26 |
+
```
|
| 27 |
+
$ python inference.py
|
| 28 |
+
Using device: cuda
|
| 29 |
+
|
| 30 |
+
Loading fine-tuned model...
|
| 31 |
+
Fine-tuned model loaded.
|
| 32 |
+
|
| 33 |
+
Loading original (pre-trained) model for comparison...
|
| 34 |
+
Original model loaded.
|
| 35 |
+
|
| 36 |
+
==================================================
|
| 37 |
+
Query: ์ผ๋ก ๋จธ์คํฌ๊ฐ ์ค๋ฆฝํ ์ ๊ธฐ์ฐจ ํ์ฌ๋ ์ด๋์ผ?
|
| 38 |
+
==================================================
|
| 39 |
+
|
| 40 |
+
--- 1. โ
Fine-tuned Model Results ---
|
| 41 |
+
Rank 1 (Score: 9.00): ํ
์ฌ๋ผ๋ ๋ชจ๋ธ S, 3, X, Y๋ฅผ ์์ฐํ๋ฉฐ ์คํ ํ์ผ๋ฟ ๊ธฐ๋ฅ์ผ๋ก ์ ๋ช
ํฉ๋๋ค.
|
| 42 |
+
Rank 2 (Score: 7.92): ์คํ์ด์คX๋ ์ฌ์ฌ์ฉ ๊ฐ๋ฅํ ๋ก์ผ์ ๊ฐ๋ฐํ์ฌ ์ฐ์ฃผ ํ์ฌ ๋น์ฉ์ ํฌ๊ฒ ๋ฎ์ท์ต๋๋ค.
|
| 43 |
+
Rank 3 (Score: 7.72): ์๋ง์กด ์น ์๋น์ค(AWS)๋ ํด๋ผ์ฐ๋ ์ปดํจํ
์์ฅ์ ์ ๋์ฃผ์์
๋๋ค.
|
| 44 |
+
Rank 4 (Score: 7.23): ์๋๊ถ ์ ์ฒ ์ ์์ธ๊ณผ ์ฃผ๋ณ ๋์๋ฅผ ์ฐ๊ฒฐํ๋ ์ค์ํ ๊ตํต์๋จ์
๋๋ค.
|
| 45 |
+
Rank 5 (Score: 5.77): ๋ํ๋ฏผ๊ตญ์ ์๋๋ ์์ธ์
๋๋ค. ์์ธ์ ๊ฒฝ์ ์ ๋ฌธํ์ ์ค์ฌ์ง์
๋๋ค.
|
| 46 |
+
Rank 6 (Score: 5.43): ์ผ๋ณธ์ ์๋๋ ๋์ฟ์
๋๋ค. ๋ฒ๊ฝ์ด ์๋ฆ๋ค์ด ๋์์ฃ .
|
| 47 |
+
Rank 7 (Score: 5.40): ํ๋์ค์ ์๋๋ ํ๋ฆฌ์ด๋ฉฐ, ์ํ ํ์ผ๋ก ์ ๋ช
ํฉ๋๋ค.
|
| 48 |
+
|
| 49 |
+
--- 2. โ Original Model Results ---
|
| 50 |
+
Rank 1 (Score: 9.13): ์๋๊ถ ์ ์ฒ ์ ์์ธ๊ณผ ์ฃผ๋ณ ๋์๋ฅผ ์ฐ๊ฒฐํ๋ ์ค์ํ ๊ตํต์๋จ์
๋๋ค.
|
| 51 |
+
Rank 2 (Score: 8.79): ํ
์ฌ๋ผ๋ ๋ชจ๋ธ S, 3, X, Y๋ฅผ ์์ฐํ๋ฉฐ ์คํ ํ์ผ๋ฟ ๊ธฐ๋ฅ์ผ๋ก ์ ๋ช
ํฉ๋๋ค.
|
| 52 |
+
Rank 3 (Score: 8.77): ์ผ๋ณธ์ ์๋๋ ๋์ฟ์
๋๋ค. ๋ฒ๊ฝ์ด ์๋ฆ๋ค์ด ๋์์ฃ .
|
| 53 |
+
Rank 4 (Score: 8.71): ๋ํ๋ฏผ๊ตญ์ ์๋๋ ์์ธ์
๋๋ค. ์์ธ์ ๊ฒฝ์ ์ ๋ฌธํ์ ์ค์ฌ์ง์
๋๋ค.
|
| 54 |
+
Rank 5 (Score: 8.53): ์๋ง์กด ์น ์๋น์ค(AWS)๋ ํด๋ผ์ฐ๋ ์ปดํจํ
์์ฅ์ ์ ๋์ฃผ์์
๋๋ค.
|
| 55 |
+
Rank 6 (Score: 8.48): ์คํ์ด์คX๋ ์ฌ์ฌ์ฉ ๊ฐ๋ฅํ ๋ก์ผ์ ๊ฐ๋ฐํ์ฌ ์ฐ์ฃผ ํ์ฌ ๋น์ฉ์ ํฌ๊ฒ ๋ฎ์ท์ต๋๋ค.
|
| 56 |
+
Rank 7 (Score: 8.24): ํ๋์ค์ ์๋๋ ํ๋ฆฌ์ด๋ฉฐ, ์ํ ํ์ผ๋ก ์ ๋ช
ํฉ๋๋ค.
|
| 57 |
+
```
|
| 58 |
|
| 59 |
**Analysis**: The fine-tuned model correctly identifies 'Tesla' by understanding the semantic relationship between the query and the document, even with no direct keyword overlap. In contrast, the original model is easily confused by distractors and fails to rank the correct answer first, demonstrating the significant impact of the ColBERT fine-tuning process.
|
| 60 |
|