| | --- |
| | language: en |
| | license: mit |
| | library_name: transformers |
| | tags: |
| | - text-classification |
| | - sentiment-analysis |
| | - fine-tuned |
| | datasets: |
| | - imdb |
| | base_model: distilbert-base-uncased |
| | --- |
| | |
| | # DistilBERT IMDb Sentiment Analysis |
| |
|
| | DistilBERT๋ฅผ IMDb ์ํ ๋ฆฌ๋ทฐ ๊ฐ์ ๋ถ์ ๋ฐ์ดํฐ์
์ผ๋ก fine-tuningํ ๋ชจ๋ธ์
๋๋ค. |
| |
|
| | ## ํ์ต ์์ธ |
| |
|
| | - **๋ฒ ์ด์ค ๋ชจ๋ธ:** distilbert-base-uncased |
| | - **๋ฐ์ดํฐ์
:** IMDb (train 2,000 / test 500 ์ํ) |
| | - **ํ์คํฌ:** ์ด์ง ๊ฐ์ ๋ถ๋ฅ (๊ธ์ /๋ถ์ ) |
| | - **Epoch:** 2 |
| | - **Learning Rate:** 2e-5 |
| | - **Batch Size:** 16 |
| | - **Max Length:** 256 |
| |
|
| | ## ์ฌ์ฉ๋ฒ |
| | ```python |
| | from transformers import pipeline |
| | |
| | classifier = pipeline("text-classification", model="AutoBot000/distilbert-imdb-sentiment") |
| | result = classifier("This movie was absolutely fantastic!") |
| | print(result) |
| | ``` |
| |
|
| | ## ํ๊ณ |
| |
|
| | - ํ์ต ๋ฐ์ดํฐ๋ฅผ 2,000๊ฐ๋ก ์ถ์ํ์ฌ ์ค์ต ๋ชฉ์ ์ผ๋ก ํ์ตํจ |
| | - ์ ์ฒด ๋ฐ์ดํฐ(25K)๋ก ํ์ต ์ ์ฑ๋ฅ ํฅ์ ๊ธฐ๋ |
| | - ์์ด ๋ฆฌ๋ทฐ ์ ์ฉ |