Notice
Recent Posts
Recent Comments
Link
์ผ | ์ | ํ | ์ | ๋ชฉ | ๊ธ | ํ |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | 6 | |
7 | 8 | 9 | 10 | 11 | 12 | 13 |
14 | 15 | 16 | 17 | 18 | 19 | 20 |
21 | 22 | 23 | 24 | 25 | 26 | 27 |
28 | 29 | 30 |
Tags
- Inductive Bias
- ์ธ์ ํ๋ ฌ
- feature map
- ์ฐ์ฐ๋ ๊ฐ์
- dp
- ์ ๊ทํ
- numpy
- bottleneck
- skip connection
- ์๋ณ์
- BFS
- ๋ฐ์ดํฐ๋ชจ๋ธ๋ง
- get_dummies()
- pytorch
- SQL
- resnet
- SQLD
- ๋ฐฑ์ค
- outer join
- 1x1 Convolution
- dfs
- Depthwise Separable Convolution
- mobilenet
- ์ํฐํฐ
- CROSS JOIN
- ์ธ์ ๋ฆฌ์คํธ
- Two Pointer
- depthwise convolution
- SQLD ํ๊ธฐ
- ๊ทธ๋ํ
Archives
- Today
- Total
๋ชฉ๋ก2024/04/23 (1)
SJ_Koding

์ด์ ๊ธ์ ์ด์ด Approch์ ๋ํ ๋ด์ฉ์ด๋ค. ์ด์ ๊ธ๊ณผ ๋ค๋ฅด๊ฒ ์ง๊ธ๋ถํฐ๋ ํต์ฌ๋ง ์์ฝํ๋ค. LLaMA-1์ Pre-training, Architecture, Optimizer, Efficient implementation์ ์ ๋ฆฌํ๋ค. LLM ๋ชจ๋ธ์์ ์ด๋ค์์ผ๋ก ๋ฐ์ดํฐ์ ์ ๊ตฌ์ถํ๊ณ , ์ผ๋งํผ์ ์์์ ์ฌ์ฉํ๋์ง, ์ด๋ค์์ผ๋ก ํ์ตํ๋์ง๋ฅผ ํ์ ํ ์ ์๋ ์น์ ์ด๋ค. ํด๋น ์น์ ์ ๋ฆฌ๋ทฐํ๋ฉด์ ๋๊ฐ LLM์ ์ ๋ฐ์ ์ธ ์ ๊ทผ๋ฐฉ์์ ํ์ ํ ์ ์์๋ค. LLaMA: Open and Efficient Foundation Language Models๋ฅผ ์์๋ณด์ - 1ํธ, Introduction ํด๋น ๋ ผ๋ฌธ์ ๋ณด๋ฉด์ LLM ์ฐ๊ตฌ์ ํฐ ํ๋ฆ์ ๋๊ฐ์ด๋ผ๋ ํ์ ํ ์ ์์๋ค. ์ต๊ทผ์ LLaMA2์ ๋นํด ๋น์ฝ์ ์ผ๋ก ์ฑ๋ฅ์ ํฅ์์ํจ (L..
LLM
2024. 4. 23. 17:21