question
stringlengths
101
2.69k
chosen
stringlengths
42
2.7k
rejected
stringlengths
42
2.18k
average curves over the 10 seeds one can discern a grokking-like delayed learning for the mixtures with relatively small amounts of AI data.
10๊ฐœ์˜ ์‹œ๋“œ์— ๋Œ€ํ•œ ํ‰๊ท  ๊ณก์„ ์—์„œ AI ๋ฐ์ดํ„ฐ์˜ ์–‘์ด ์ƒ๋Œ€์ ์œผ๋กœ ์ ์€ ํ˜ผํ•ฉ๋ฌผ์— ๋Œ€ํ•ด grokking๊ณผ ์œ ์‚ฌํ•œ ์ง€์—ฐ ํ•™์Šต์„ ์‹๋ณ„ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
10๊ฐœ์˜ ์‹œ๋“œ์— ๋Œ€ํ•œ ํ‰๊ท  ๊ณก์„ ์—์„œ AI ๋ฐ์ดํ„ฐ์˜ ์–‘์ด ์ƒ๋Œ€์ ์œผ๋กœ ์ ์€ ํ˜ผํ•ฉ๋ฌผ์— ๋Œ€ํ•ด ๊ทธ๋กœํ‚น๊ณผ ์œ ์‚ฌํ•œ ์ง€์—ฐ ํ•™์Šต์„ ์‹๋ณ„ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Therefore, the transformation \(p\to p_{\text{AI}}\) amounts to chopping off the tail of \(p\) at rank \(k\), where \(k\) is as given above.
๋”ฐ๋ผ์„œ \(p\to p_{\text{AI}}\)์˜ ๋ณ€ํ™˜์€ \(p\)์˜ ๊ผฌ๋ฆฌ๋ฅผ ์ˆœ์œ„ \(k\)์—์„œ ์ž˜๋ผ๋‚ด๋Š” ๊ฒƒ๊ณผ ๊ฐ™์œผ๋ฉฐ, ์—ฌ๊ธฐ์„œ \(k\)๋Š” ์œ„์— ์ฃผ์–ด์ง„ ๋Œ€๋กœ์ž…๋‹ˆ๋‹ค.
๋”ฐ๋ผ์„œ \(p\to p_{\text{AI}}\)์˜ ๋ณ€ํ™˜์€ \(p\)์˜ ๊ผฌ๋ฆฌ๋ฅผ ์ˆœ์œ„ \(k\)์—์„œ ์ž˜๋ผ๋‚ด๋Š” ๊ฒƒ๊ณผ ๊ฐ™์œผ๋ฉฐ, ์—ฌ๊ธฐ์„œ \(k\)๋Š” ์œ„์— ์ฃผ์–ด์ง„ ๋Œ€๋กœ์ž…๋‹ˆ๋‹ค.
Figure 1: Training stages with parameter freezing. The fire and snowflake emojis indicate the trainable and frozen parameters respectively.
๊ทธ๋ฆผ 1: ํŒŒ๋ผ๋ฏธํ„ฐ ๋™๊ฒฐ์ด ์žˆ๋Š” ํ›ˆ๋ จ ๋‹จ๊ณ„. ๋ถˆ๊ฝƒ๊ณผ ๋ˆˆ์†ก์ด ์ด๋ชจํ‹ฐ์ฝ˜์€ ๊ฐ๊ฐ ํ›ˆ๋ จ ๊ฐ€๋Šฅํ•œ ํŒŒ๋ผ๋ฏธํ„ฐ์™€ ๋™๊ฒฐ๋œ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ๋‚˜ํƒ€๋ƒ…๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 1: ํŒŒ๋ผ๋ฏธํ„ฐ ๋™๊ฒฐ์ด ์žˆ๋Š” ํ›ˆ๋ จ ๋‹จ๊ณ„. ๋ถˆ๊ฝƒ๊ณผ ๋ˆˆ์†ก์ด ์ด๋ชจํ‹ฐ์ฝ˜์€ ๊ฐ๊ฐ ํ›ˆ๋ จ ๊ฐ€๋Šฅํ•œ ํŒŒ๋ผ๋ฏธํ„ฐ์™€ ๋™๊ฒฐ๋œ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ๋‚˜ํƒ€๋ƒ…๋‹ˆ๋‹ค.
Throughout the whole process, a validator checks the sanity of the data and gives detailed real-time feedback on the work of the annotator.
์ „์ฒด ํ”„๋กœ์„ธ์Šค์—์„œ ๊ฒ€์ฆ์ž๋Š” ๋ฐ์ดํ„ฐ์˜ ๊ฑด์ „์„ฑ์„ ํ™•์ธํ•˜๊ณ  ์ฃผ์„ ์ž‘์„ฑ์ž์˜ ์ž‘์—…์— ๋Œ€ํ•œ ์ƒ์„ธํ•œ ์‹ค์‹œ๊ฐ„ ํ”ผ๋“œ๋ฐฑ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
์ „์ฒด ํ”„๋กœ์„ธ์Šค์—์„œ ๊ฒ€์ฆ์ž๋Š” ๋ฐ์ดํ„ฐ์˜ ๊ฑด์ „์„ฑ์„ ํ™•์ธํ•˜๊ณ  ์ฃผ์„ ์ž‘์„ฑ์ž์˜ ์ž‘์—…์— ๋Œ€ํ•œ ์ƒ์„ธํ•œ ์‹ค์‹œ๊ฐ„ ํ”ผ๋“œ๋ฐฑ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
refuse benign prompts [see 31, 68]. These limitations will be addressed in future versions of the benchmark by expanding our scope of work.
์•…์˜์  ํ”„๋กฌํ”„ํŠธ๋ฅผ ๊ฑฐ๋ถ€ํ•ฉ๋‹ˆ๋‹ค[31, 68 ์ฐธ์กฐ]. ์ด๋Ÿฌํ•œ ํ•œ๊ณ„๋Š” ํ–ฅํ›„ ๋ฒค์น˜๋งˆํฌ ๋ฒ„์ „์—์„œ ์ž‘์—… ๋ฒ”์œ„๋ฅผ ํ™•์žฅํ•˜์—ฌ ํ•ด๊ฒฐํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
์•…์˜์  ํ”„๋กฌํ”„ํŠธ๋ฅผ ๊ฑฐ๋ถ€ํ•ฉ๋‹ˆ๋‹ค[31, 68 ์ฐธ์กฐ]. ์ด๋Ÿฌํ•œ ํ•œ๊ณ„๋Š” ํ–ฅํ›„ ๋ฒค์น˜๋งˆํฌ ๋ฒ„์ „์—์„œ ์ž‘์—… ๋ฒ”์œ„๋ฅผ ํ™•์žฅํ•˜์—ฌ ํ•ด๊ฒฐํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
2. There are 35 fewer rubies than diamonds. Since there are 175 diamonds, we can calculate the number of rubies by subtracting 35 from 175.
2. ๋‹ค์ด์•„๋ชฌ๋“œ๋ณด๋‹ค ๋ฃจ๋น„๊ฐ€ 35๊ฐœ ์ ์Šต๋‹ˆ๋‹ค. ๋‹ค์ด์•„๋ชฌ๋“œ๊ฐ€ 175๊ฐœ์ด๋ฏ€๋กœ 175์—์„œ 35๋ฅผ ๋นผ๋ฉด ๋ฃจ๋น„์˜ ๊ฐœ์ˆ˜๋ฅผ ๊ณ„์‚ฐํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
2. ๋‹ค์ด์•„๋ชฌ๋“œ๋ณด๋‹ค ๋ฃจ๋น„๊ฐ€ 35๊ฐœ ์ ์Šต๋‹ˆ๋‹ค. ๋‹ค์ด์•„๋ชฌ๋“œ๊ฐ€ 175๊ฐœ์ด๋ฏ€๋กœ 175์—์„œ 35๋ฅผ ๋นผ๋ฉด ๋ฃจ๋น„์˜ ๊ฐœ์ˆ˜๋ฅผ ๊ณ„์‚ฐํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Footnote 4: Following Lightman et al. (2023), we utilize a subset of 500 test samples from the MATH benchmark for experimental efficiency.
๊ฐ์ฃผ 4: Lightman ๋“ฑ(2023)์— ๋”ฐ๋ผ ์‹คํ—˜ ํšจ์œจ์„ฑ์„ ์œ„ํ•ด MATH ๋ฒค์น˜๋งˆํฌ์˜ ํ…Œ์ŠคํŠธ ์ƒ˜ํ”Œ 500๊ฐœ ํ•˜์œ„ ์ง‘ํ•ฉ์„ ํ™œ์šฉํ•ฉ๋‹ˆ๋‹ค.
๊ฐ์ฃผ 4: Lightman ๋“ฑ(2023)์— ๋”ฐ๋ผ ์‹คํ—˜ ํšจ์œจ์„ฑ์„ ์œ„ํ•ด ์ˆ˜ํ•™ ๋ฒค์น˜๋งˆํฌ์˜ ํ…Œ์ŠคํŠธ ์ƒ˜ํ”Œ 500๊ฐœ ํ•˜์œ„ ์ง‘ํ•ฉ์„ ํ™œ์šฉํ•ฉ๋‹ˆ๋‹ค.
In this work, we address this gap and conduct an extensive study in which we measure the impact of the tokenizer on the model performance.
์ด ์ž‘์—…์—์„œ๋Š” ์ด๋Ÿฌํ•œ ๊ฒฉ์ฐจ๋ฅผ ํ•ด์†Œํ•˜๊ณ  tokenizer๊ฐ€ ๋ชจ๋ธ ์„ฑ๋Šฅ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ์ธก์ •ํ•˜๋Š” ๊ด‘๋ฒ”์œ„ํ•œ ์—ฐ๊ตฌ๋ฅผ ์ˆ˜ํ–‰ํ•ฉ๋‹ˆ๋‹ค.
์ด ์ž‘์—…์—์„œ๋Š” ์ด๋Ÿฌํ•œ ๊ฒฉ์ฐจ๋ฅผ ํ•ด์†Œํ•˜๊ณ  ํ† ํฐํ™”๊ธฐ๊ฐ€ ๋ชจ๋ธ ์„ฑ๋Šฅ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ์ธก์ •ํ•˜๋Š” ๊ด‘๋ฒ”์œ„ํ•œ ์—ฐ๊ตฌ๋ฅผ ์ˆ˜ํ–‰ํ•ฉ๋‹ˆ๋‹ค.
Figure 12 \(|\) R-GSM example where the original problem can be correctly solved by GPT-4 Turbo, but the model fails on the reordered one.
๊ทธ๋ฆผ 12 \(|\) R-GSM ์˜ˆ์ œ์—์„œ ์›๋ž˜ ๋ฌธ์ œ๋Š” GPT-4 Turbo๋กœ ์ •ํ™•ํ•˜๊ฒŒ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ์ง€๋งŒ, ์ˆœ์„œ๊ฐ€ ๋ฐ”๋€ ๋ฌธ์ œ์—์„œ๋Š” ๋ชจ๋ธ์ด ์‹คํŒจํ•ฉ๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 12 \(|\) R-GSM ์˜ˆ์ œ์—์„œ ์›๋ž˜ ๋ฌธ์ œ๋Š” GPT-4 ํ„ฐ๋ณด๋กœ ์ •ํ™•ํ•˜๊ฒŒ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ์ง€๋งŒ, ์ˆœ์„œ๊ฐ€ ๋ฐ”๋€ ๋ฌธ์ œ์—์„œ๋Š” ๋ชจ๋ธ์ด ์‹คํŒจํ•ฉ๋‹ˆ๋‹ค.
PersonU starts the conversation and only asks questions. Most of PersonU's questions are follow-up or continuations of previous questions.
PersonU๋Š” ๋Œ€ํ™”๋ฅผ ์‹œ์ž‘ํ•˜๊ณ  ์งˆ๋ฌธ๋งŒ ํ•ฉ๋‹ˆ๋‹ค. PersonU์˜ ์งˆ๋ฌธ์€ ๋Œ€๋ถ€๋ถ„ ์ด์ „ ์งˆ๋ฌธ์˜ ํ›„์† ์งˆ๋ฌธ ๋˜๋Š” ์ด์–ด์„œ ํ•˜๋Š” ์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค.
PersonU๋Š” ๋Œ€ํ™”๋ฅผ ์‹œ์ž‘ํ•˜๊ณ  ์งˆ๋ฌธ๋งŒ ํ•ฉ๋‹ˆ๋‹ค. PersonU์˜ ์งˆ๋ฌธ์€ ๋Œ€๋ถ€๋ถ„ ์ด์ „ ์งˆ๋ฌธ์˜ ํ›„์† ์งˆ๋ฌธ ๋˜๋Š” ์ด์–ด์„œ ํ•˜๋Š” ์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค.
7. Now, we have another instance of \(f\). Since \(f\) is odd, we have \(f(-x)=-f(x)\). So, we can rewrite the expression as \(-f(f(x))\).
7. ์ด์ œ \(f\)์˜ ๋˜ ๋‹ค๋ฅธ ์ธ์Šคํ„ด์Šค๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. \(f\)๋Š” ํ™€์ˆ˜์ด๋ฏ€๋กœ \(f(-x)=-f(x)\)๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹์„ \(-f(f(x))\)๋กœ ๋‹ค์‹œ ์“ธ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
7. ์ด์ œ \(f\)์˜ ๋˜ ๋‹ค๋ฅธ ์ธ์Šคํ„ด์Šค๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. f\)๋Š” ํ™€์ˆ˜์ด๋ฏ€๋กœ \(f(-x)=-f(x)\)๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹์„ \(-f(f(x))\)๋กœ ๋‹ค์‹œ ์“ธ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Contributors to **license creation and risk assessment** include David Atkinson, Jesse Dodge, Jennifer Dumas, Crystal Nam, and Will Smith.
๋ผ์ด์„ ์Šค ์ƒ์„ฑ ๋ฐ ์œ„ํ—˜ ํ‰๊ฐ€**์— ๊ธฐ์—ฌํ•œ ์‚ฌ๋žŒ์œผ๋กœ๋Š” David Atkinson, Jesse Dodge, Jennifer Dumas, Crystal Nam, Will Smith ๋“ฑ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
๋ผ์ด์„ ์Šค ์ƒ์„ฑ ๋ฐ ์œ„ํ—˜ ํ‰๊ฐ€**์— ๊ธฐ์—ฌํ•œ ์‚ฌ๋žŒ์œผ๋กœ๋Š” David Atkinson, Jesse Dodge, Jennifer Dumas, Crystal Nam, Will Smith ๋“ฑ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
summary on this rapidly developing field is of great significance to comprehensively understand it and benefit to inspire future research.
์ด ๋น ๋ฅด๊ฒŒ ๋ฐœ์ „ํ•˜๋Š” ๋ถ„์•ผ์— ๋Œ€ํ•œ ์š”์•ฝ์€ ์ด๋ฅผ ์ข…ํ•ฉ์ ์œผ๋กœ ์ดํ•ดํ•˜๊ณ  ํ–ฅํ›„ ์—ฐ๊ตฌ์— ์˜๊ฐ์„ ๋ถˆ์–ด๋„ฃ๋Š” ๋ฐ ํฐ ์˜๋ฏธ๊ฐ€ ์žˆ๋‹ค.
์ด ๋น ๋ฅด๊ฒŒ ๋ฐœ์ „ํ•˜๋Š” ๋ถ„์•ผ์— ๋Œ€ํ•œ ์š”์•ฝ์€ ์ด๋ฅผ ์ข…ํ•ฉ์ ์œผ๋กœ ์ดํ•ดํ•˜๊ณ  ํ–ฅํ›„ ์—ฐ๊ตฌ์— ์˜๊ฐ์„ ๋ถˆ์–ด๋„ฃ๋Š” ๋ฐ ํฐ ์˜๋ฏธ๊ฐ€ ์žˆ๋‹ค.
Since the logarithm gives an exact integer, to make the question valid, we will modify it to use a number that is not an exact power of 2.
๋กœ๊ทธ๊ฐ€ ์ •ํ™•ํ•œ ์ •์ˆ˜๋ฅผ ์ œ๊ณตํ•˜๋ฏ€๋กœ ๋ฌธ์ œ์˜ ์œ ํšจ์„ฑ์„ ๋†’์ด๊ธฐ ์œ„ํ•ด 2์˜ ์ •ํ™•ํ•œ ๊ฑฐ๋“ญ์ œ๊ณฑ์ด ์•„๋‹Œ ์ˆซ์ž๋ฅผ ์‚ฌ์šฉํ•˜๋„๋ก ์ˆ˜์ •ํ•˜๊ฒ ์Šต๋‹ˆ๋‹ค.
๋กœ๊ทธ๊ฐ€ ์ •ํ™•ํ•œ ์ •์ˆ˜๋ฅผ ์ œ๊ณตํ•˜๋ฏ€๋กœ ๋ฌธ์ œ์˜ ์œ ํšจ์„ฑ์„ ๋†’์ด๊ธฐ ์œ„ํ•ด 2์˜ ์ •ํ™•ํ•œ ๊ฑฐ๋“ญ์ œ๊ณฑ์ด ์•„๋‹Œ ์ˆซ์ž๋ฅผ ์‚ฌ์šฉํ•˜๋„๋ก ์ˆ˜์ •ํ•˜๊ฒ ์Šต๋‹ˆ๋‹ค.
- The calculated probability should not be excessively confident in either direction due to the uncertainty expressed by McCarthy himself.
- McCarthy๊ฐ€ ์ง์ ‘ ํ‘œํ˜„ํ•œ ๋ถˆํ™•์‹ค์„ฑ์œผ๋กœ ์ธํ•ด ๊ณ„์‚ฐ๋œ ํ™•๋ฅ ์ด ์–ด๋Š ํ•œ ๋ฐฉํ–ฅ์œผ๋กœ ๊ณผ๋„ํ•˜๊ฒŒ ํ™•์‹ ํ•ด์„œ๋Š” ์•ˆ ๋œ๋‹ค.
- ๋งฅ์นด์‹œ๊ฐ€ ์ง์ ‘ ํ‘œํ˜„ํ•œ ๋ถˆํ™•์‹ค์„ฑ์œผ๋กœ ์ธํ•ด ๊ณ„์‚ฐ๋œ ํ™•๋ฅ ์ด ์–ด๋Š ํ•œ ๋ฐฉํ–ฅ์œผ๋กœ ๊ณผ๋„ํ•˜๊ฒŒ ํ™•์‹ ํ•ด์„œ๋Š” ์•ˆ ๋œ๋‹ค.
Cabannes et al. (2023) proposed that iid random embeddings from the uniform distribution on the unit-sphere in \(\mathbb{R}^{d}\) be used.
Cabannes ๋“ฑ(2023)์€ \(\mathbb{R}^{d}\)์—์„œ ๋‹จ์œ„ ๊ตฌ์˜ ๊ท ๋“ฑ ๋ถ„ํฌ์—์„œ iid ๋žœ๋ค embedding์„ ์‚ฌ์šฉํ•  ๊ฒƒ์„ ์ œ์•ˆํ–ˆ๋‹ค.
Cabannes ๋“ฑ(2023)์€ \(\mathbb{R}^{d}\)์—์„œ ๋‹จ์œ„ ๊ตฌ์˜ ๊ท ๋“ฑ ๋ถ„ํฌ์—์„œ iid ๋žœ๋ค ์ž„๋ฒ ๋”ฉ์„ ์‚ฌ์šฉํ•  ๊ฒƒ์„ ์ œ์•ˆํ–ˆ๋‹ค.
The annotators were all trained in safety annotation and experienced in undertaking such work, with between 3 and 5 years experience each.
์ฃผ์„ ์ž‘์„ฑ์ž๋Š” ๋ชจ๋‘ ์•ˆ์ „ ์ฃผ์„ ๊ต์œก์„ ๋ฐ›์•˜์œผ๋ฉฐ, 3~5๋…„์˜ ๊ฒฝ๋ ฅ์„ ๊ฐ€์ง„ ์•ˆ์ „ ์ฃผ์„ ์ž‘์„ฑ ๊ฒฝํ—˜์ด ์žˆ๋Š” ์‚ฌ๋žŒ๋“ค์ž…๋‹ˆ๋‹ค.
์ฃผ์„ ์ž‘์„ฑ์ž๋Š” ๋ชจ๋‘ ์•ˆ์ „ ์ฃผ์„ ๊ต์œก์„ ๋ฐ›์•˜์œผ๋ฉฐ, 3~5๋…„์˜ ๊ฒฝ๋ ฅ์„ ๊ฐ€์ง„ ์•ˆ์ „ ์ฃผ์„ ์ž‘์„ฑ ๊ฒฝํ—˜์ด ์žˆ๋Š” ์‚ฌ๋žŒ๋“ค์ž…๋‹ˆ๋‹ค.
Figure 11: Examples of claims accompanied by annotator labels and reasoning, along with predictions made by Claude-3-Opus and GPT-4-Turbo.
๊ทธ๋ฆผ 11: ์ฃผ์„๊ฐ€ ๋ ˆ์ด๋ธ”๊ณผ ์ถ”๋ก ์ด ํฌํ•จ๋œ ํด๋ ˆ์ž„์˜ ์˜ˆ์™€ Claude-3-Opus ๋ฐ GPT-4-Turbo์˜ ์˜ˆ์ธก ๊ฒฐ๊ณผ.
๊ทธ๋ฆผ 11: ์ฃผ์„๊ฐ€ ๋ ˆ์ด๋ธ”๊ณผ ์ถ”๋ก ์ด ํฌํ•จ๋œ ํด๋ ˆ์ž„์˜ ์˜ˆ์™€ Claude-3-Opus ๋ฐ GPT-4-Turbo์˜ ์˜ˆ์ธก ๊ฒฐ๊ณผ.
- Demonstrating leadership: High. As a former president, Trump might want to assert his dominance in the Republican field from the outset.
- ๋ฆฌ๋”์‹ญ ์ž…์ฆ: ๋†’์Œ. ํŠธ๋Ÿผํ”„๋Š” ์ „ ๋Œ€ํ†ต๋ น์œผ๋กœ์„œ ๊ณตํ™”๋‹น ๊ฒฝ์„ ์—์„œ ์ฒ˜์Œ๋ถ€ํ„ฐ ์ž์‹ ์˜ ์šฐ์œ„๋ฅผ ์ฃผ์žฅํ•˜๊ณ  ์‹ถ์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ๋ฆฌ๋”์‹ญ ์ž…์ฆ: ๋†’์Œ. ํŠธ๋Ÿผํ”„๋Š” ์ „ ๋Œ€ํ†ต๋ น์œผ๋กœ์„œ ๊ณตํ™”๋‹น ๊ฒฝ์„ ์—์„œ ์ฒ˜์Œ๋ถ€ํ„ฐ ์ž์‹ ์˜ ์šฐ์œ„๋ฅผ ์ฃผ์žฅํ•˜๊ณ  ์‹ถ์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
To find the value of \(f(-2)+f(-1)+f(0)\), we need to evaluate the function \(f(x)\) at each of the given points and then sum the results.
\(f(-2)+f(-1)+f(0)\)์˜ ๊ฐ’์„ ๊ตฌํ•˜๋ ค๋ฉด ์ฃผ์–ด์ง„ ๊ฐ ์ง€์ ์—์„œ ํ•จ์ˆ˜ \(f(x)\)๋ฅผ ํ‰๊ฐ€ํ•œ ๋‹ค์Œ ๊ฒฐ๊ณผ๋ฅผ ํ•ฉ์‚ฐํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
f(-2)+f(-1)+f(0)\)์˜ ๊ฐ’์„ ๊ตฌํ•˜๋ ค๋ฉด ์ฃผ์–ด์ง„ ๊ฐ ์ง€์ ์—์„œ ํ•จ์ˆ˜ \(f(x)\)๋ฅผ ํ‰๊ฐ€ํ•œ ๋‹ค์Œ ๊ฒฐ๊ณผ๋ฅผ ํ•ฉ์‚ฐํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
Motivated by this, we compare validation sets side-by-side (in terms of distance to training set) in Figure 5, and we see a similar trend.
์ด์— ๋™๊ธฐ๋ฅผ ๋ถ€์—ฌํ•˜์—ฌ ๊ทธ๋ฆผ 5์—์„œ validation ์„ธํŠธ๋ฅผ ๋‚˜๋ž€ํžˆ ๋น„๊ต(ํ›ˆ๋ จ ์„ธํŠธ์™€์˜ ๊ฑฐ๋ฆฌ ์ธก๋ฉด)ํ•œ ๊ฒฐ๊ณผ, ์œ ์‚ฌํ•œ ๊ฒฝํ–ฅ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.
์ด์— ๋™๊ธฐ๋ฅผ ๋ถ€์—ฌํ•˜์—ฌ ๊ทธ๋ฆผ 5์—์„œ ์œ ์‚ฌ๋„ ์„ธํŠธ๋ฅผ ๋‚˜๋ž€ํžˆ ๋น„๊ต(ํ›ˆ๋ จ ์„ธํŠธ์™€์˜ ๊ฑฐ๋ฆฌ ์ธก๋ฉด)ํ•œ ๊ฒฐ๊ณผ, ์œ ์‚ฌํ•œ ๊ฒฝํ–ฅ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.
Footnote 5: Both orders will often appear in the same document. For example: โ€œValentina Tereshkova was the first woman to travel to space.
๊ฐ์ฃผ 5: ๋‘ ๋ช…๋ น์€ ์ข…์ข… ๊ฐ™์€ ๋ฌธ์„œ์— ํ‘œ์‹œ๋ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด "๋ฐœ๋ Œํ‹ฐ๋‚˜ ํ…Œ๋ ˆ์Šˆ์ฝ”๋ฐ”๋Š” ์ตœ์ดˆ๋กœ ์šฐ์ฃผ๋ฅผ ์—ฌํ–‰ํ•œ ์—ฌ์„ฑ์ž…๋‹ˆ๋‹ค.
๊ฐ์ฃผ 5: ๋‘ ๋ช…๋ น์€ ์ข…์ข… ๊ฐ™์€ ๋ฌธ์„œ์— ํ‘œ์‹œ๋ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด "๋ฐœ๋ Œํ‹ฐ๋‚˜ ํ…Œ๋ ˆ์Šˆ์ฝ”๋ฐ”๋Š” ์ตœ์ดˆ๋กœ ์šฐ์ฃผ๋ฅผ ์—ฌํ–‰ํ•œ ์—ฌ์„ฑ์ž…๋‹ˆ๋‹ค.
To ensure the question can be solved, we need to check if the logarithm given can be evaluated and falls between two consecutive integers.
์งˆ๋ฌธ์„ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ๋Š”์ง€ ํ™•์ธํ•˜๋ ค๋ฉด ์ฃผ์–ด์ง„ logarithm์„ ํ‰๊ฐ€ํ•  ์ˆ˜ ์žˆ๋Š”์ง€, ๊ทธ๋ฆฌ๊ณ  ์—ฐ์†๋œ ๋‘ ์ •์ˆ˜ ์‚ฌ์ด์— ์žˆ๋Š”์ง€ ํ™•์ธํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
์งˆ๋ฌธ์„ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ๋Š”์ง€ ํ™•์ธํ•˜๋ ค๋ฉด ์ฃผ์–ด์ง„ ๋กœ๊ทธ๋ฅผ ํ‰๊ฐ€ํ•  ์ˆ˜ ์žˆ๋Š”์ง€, ๊ทธ๋ฆฌ๊ณ  ์—ฐ์†๋œ ๋‘ ์ •์ˆ˜ ์‚ฌ์ด์— ์žˆ๋Š”์ง€ ํ™•์ธํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
- Engaging with opponents: High. Trump is known for his confrontational style, which could play out to his advantage in a debate setting.
- ์ƒ๋Œ€๋ฐฉ๊ณผ์˜ ๊ต์ „: ๋†’์Œ. ํŠธ๋Ÿผํ”„๋Š” ๋Œ€๋ฆฝ์ ์ธ ์Šคํƒ€์ผ๋กœ ์œ ๋ช…ํ•˜๋ฉฐ, ์ด๋Š” ํ† ๋ก ํšŒ์—์„œ ์œ ๋ฆฌํ•˜๊ฒŒ ์ž‘์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ์ƒ๋Œ€๋ฐฉ๊ณผ์˜ ๊ต์ „: ๋†’์Œ. ํŠธ๋Ÿผํ”„๋Š” ๋Œ€๋ฆฝ์ ์ธ ์Šคํƒ€์ผ๋กœ ์œ ๋ช…ํ•˜๋ฉฐ, ์ด๋Š” ํ† ๋ก ํšŒ์—์„œ ์œ ๋ฆฌํ•˜๊ฒŒ ์ž‘์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
PersonU starts the conversation and only asks questions. Most of PersonU's questionsare follow-up or continuations of previous questions.
PersonU๋Š” ๋Œ€ํ™”๋ฅผ ์‹œ์ž‘ํ•˜๊ณ  ์งˆ๋ฌธ๋งŒ ํ•ฉ๋‹ˆ๋‹ค. PersonU์˜ ๋Œ€๋ถ€๋ถ„์˜ ์งˆ๋ฌธ์€ ์ด์ „ ์งˆ๋ฌธ์˜ ํ›„์† ์งˆ๋ฌธ ๋˜๋Š” ์ด์–ด์„œ ํ•˜๋Š” ์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค.
PersonU๋Š” ๋Œ€ํ™”๋ฅผ ์‹œ์ž‘ํ•˜๊ณ  ์งˆ๋ฌธ๋งŒ ํ•ฉ๋‹ˆ๋‹ค. PersonU์˜ ๋Œ€๋ถ€๋ถ„์˜ ์งˆ๋ฌธ์€ ์ด์ „ ์งˆ๋ฌธ์˜ ํ›„์† ์งˆ๋ฌธ ๋˜๋Š” ์ด์–ด์„œ ํ•˜๋Š” ์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค.
On Thursday, the mechanic repaired 6 truck tires and 4 car tires. The mechanic charges $60 for each truck tire and $40 for each car tire.
๋ชฉ์š”์ผ์— ์ •๋น„์‚ฌ๋Š” ํŠธ๋Ÿญ ํƒ€์ด์–ด 6๊ฐœ์™€ ์Šน์šฉ์ฐจ ํƒ€์ด์–ด 4๊ฐœ๋ฅผ ์ˆ˜๋ฆฌํ–ˆ์Šต๋‹ˆ๋‹ค. ์ •๋น„์‚ฌ๋Š” ํŠธ๋Ÿญ ํƒ€์ด์–ด ํ•œ ๊ฐœ๋‹น 60๋‹ฌ๋Ÿฌ, ์Šน์šฉ์ฐจ ํƒ€์ด์–ด ํ•œ ๊ฐœ๋‹น 40๋‹ฌ๋Ÿฌ๋ฅผ ์ฒญ๊ตฌํ•ฉ๋‹ˆ๋‹ค.
๋ชฉ์š”์ผ์— ์ •๋น„์‚ฌ๋Š” ํŠธ๋Ÿญ ํƒ€์ด์–ด 6๊ฐœ์™€ ์Šน์šฉ์ฐจ ํƒ€์ด์–ด 4๊ฐœ๋ฅผ ์ˆ˜๋ฆฌํ–ˆ์Šต๋‹ˆ๋‹ค. ์ •๋น„์‚ฌ๋Š” ํŠธ๋Ÿญ ํƒ€์ด์–ด ํ•œ ๊ฐœ๋‹น 60๋‹ฌ๋Ÿฌ, ์Šน์šฉ์ฐจ ํƒ€์ด์–ด ํ•œ ๊ฐœ๋‹น 40๋‹ฌ๋Ÿฌ๋ฅผ ์ฒญ๊ตฌํ•ฉ๋‹ˆ๋‹ค.
Figure 1: Comparative performance analysis of Baichuan-7B, DeepSeek-7B, and DeepSeek-67B models across MMLU, CMMLU, and CEval benchmarks.
๊ทธ๋ฆผ 1: MMLU, CMMLU ๋ฐ CEval ๋ฒค์น˜๋งˆํฌ์—์„œ Baichuan-7B, DeepSeek-7B ๋ฐ DeepSeek-67B ๋ชจ๋ธ์˜ ๋น„๊ต ์„ฑ๋Šฅ ๋ถ„์„.
๊ทธ๋ฆผ 1: MMLU, CMMLU ๋ฐ CEval ๋ฒค์น˜๋งˆํฌ์—์„œ Baichuan-7B, DeepSeek-7B ๋ฐ DeepSeek-67B ๋ชจ๋ธ์˜ ๋น„๊ต ์„ฑ๋Šฅ ๋ถ„์„.
In this section, we present an overview about the background of LLMs and then summarize the technical evolution of the GPT-series models.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ๋ฐฐ๊ฒฝ์— ๋Œ€ํ•œ ๊ฐœ์š”๋ฅผ ์ œ์‹œํ•œ ๋‹ค์Œ GPT ์‹œ๋ฆฌ์ฆˆ ๋ชจ๋ธ์˜ ๊ธฐ์ˆ ์  ์ง„ํ™”๋ฅผ ์š”์•ฝํ•œ๋‹ค.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ๋ฐฐ๊ฒฝ์— ๋Œ€ํ•œ ๊ฐœ์š”๋ฅผ ์ œ์‹œํ•œ ๋‹ค์Œ GPT ์‹œ๋ฆฌ์ฆˆ ๋ชจ๋ธ์˜ ๊ธฐ์ˆ ์  ์ง„ํ™”๋ฅผ ์š”์•ฝํ•œ๋‹ค.
Footnote 10: We use \(\mathbb{E}_{n}\) or \(\mathbb{E}_{n,a}\) to denote uniform random selection of \(n\in\mathcal{N},a\in\mathcal{A}\).
์ฃผ 10: \(\mathbb{E}_{n}\) ๋˜๋Š” \(\mathbb{E}_{n,a}\)๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ \(n\in\mathcal{N},a\in\mathcal{A}\)์˜ ๊ท ์ผํ•œ ๋ฌด์ž‘์œ„ ์„ ํƒ์„ ๋‚˜ํƒ€๋ƒ…๋‹ˆ๋‹ค.
์ฃผ 10: \(\mathbb{E}_{n}\) ๋˜๋Š” \(\mathbb{E}_{n,a}\)๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ \(n\in\mathcal{N},a\in\mathcal{A}\)์˜ ๊ท ์ผํ•œ ๋ฌด์ž‘์œ„ ์„ ํƒ์„ ๋‚˜ํƒ€๋ƒ…๋‹ˆ๋‹ค.
**Does the dataset contain data that, if viewed directly, might be offensive, insulting, threatening, or might otherwise cause anxiety?**
**๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์— ์ง์ ‘ ๋ณด๊ณ  ์žˆ์œผ๋ฉด ๋ถˆ์พŒ๊ฐ, ๋ชจ์š•๊ฐ, ์œ„ํ˜‘์„ ๋Š๋ผ๊ฑฐ๋‚˜ ๋ถˆ์•ˆ์„ ์œ ๋ฐœํ•  ์ˆ˜ ์žˆ๋Š” ๋ฐ์ดํ„ฐ๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ๋‚˜์š”?**
**๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์— ์ง์ ‘ ๋ณด๊ณ  ์žˆ์œผ๋ฉด ๋ถˆ์พŒ๊ฐ, ๋ชจ์š•๊ฐ, ์œ„ํ˜‘์„ ๋Š๋ผ๊ฑฐ๋‚˜ ๋ถˆ์•ˆ์„ ์œ ๋ฐœํ•  ์ˆ˜ ์žˆ๋Š” ๋ฐ์ดํ„ฐ๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ๋‚˜์š”?**
The specific interfaces used for the mturk HITs are provided in Appendix B.2. Statistics of the constructed datasets are in Appendix B.3.
mturk HIT์— ์‚ฌ์šฉ๋œ ๊ตฌ์ฒด์ ์ธ ์ธํ„ฐํŽ˜์ด์Šค๋Š” ๋ถ€๋ก B.2์— ๋‚˜์™€ ์žˆ๋‹ค. ๊ตฌ์„ฑ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ํ†ต๊ณ„๋Š” ๋ถ€๋ก B.3์— ๋‚˜์™€ ์žˆ๋‹ค.
mturk HIT์— ์‚ฌ์šฉ๋œ ๊ตฌ์ฒด์ ์ธ ์ธํ„ฐํŽ˜์ด์Šค๋Š” ๋ถ€๋ก B.2์— ๋‚˜์™€ ์žˆ๋‹ค. ๊ตฌ์„ฑ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ํ†ต๊ณ„๋Š” ๋ถ€๋ก B.3์— ๋‚˜์™€ ์žˆ๋‹ค.
Footnote 6: A summable series of nonnegative numbers (like in \(a_{T}(i)\) and \(b_{T}(i)\)) can be reordered without changing the value.
์ฃผ 6: ์Œ์ˆ˜๊ฐ€ ์•„๋‹Œ ์ˆ˜์˜ ํ•ฉ์‚ฐ ๊ฐ€๋Šฅํ•œ ๊ธ‰์ˆ˜(์˜ˆ: \(a_{T}(i)\) ๋ฐ \(b_{T}(i)\))๋Š” ๊ฐ’์„ ๋ณ€๊ฒฝํ•˜์ง€ ์•Š๊ณ  ์ˆœ์„œ๋ฅผ ๋ฐ”๊ฟ€ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
์ฃผ 6: ์Œ์ˆ˜๊ฐ€ ์•„๋‹Œ ์ˆ˜์˜ ํ•ฉ์‚ฐ ๊ฐ€๋Šฅํ•œ ๊ธ‰์ˆ˜(์˜ˆ: \(a_{T}(i)\) ๋ฐ \(b_{T}(i)\))๋Š” ๊ฐ’์„ ๋ณ€๊ฒฝํ•˜์ง€ ์•Š๊ณ  ์ˆœ์„œ๋ฅผ ๋ฐ”๊ฟ€ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
(3) Reasoning: Assessing the model's ability to execute correct reasoning processes or devise valid reasoning concepts to solve problems.
(3) Reasoning: ๋ชจ๋ธ์ด ์˜ฌ๋ฐ”๋ฅธ ์ถ”๋ก  ํ”„๋กœ์„ธ์Šค๋ฅผ ์‹คํ–‰ํ•˜๊ฑฐ๋‚˜ ๋ฌธ์ œ ํ•ด๊ฒฐ์„ ์œ„ํ•œ ์œ ํšจํ•œ ์ถ”๋ก  ๊ฐœ๋…์„ ๊ณ ์•ˆํ•  ์ˆ˜ ์žˆ๋Š” ๋Šฅ๋ ฅ์„ ํ‰๊ฐ€ํ•ฉ๋‹ˆ๋‹ค.
(3) ์ถ”๋ก : ๋ชจ๋ธ์ด ์˜ฌ๋ฐ”๋ฅธ ์ถ”๋ก  ํ”„๋กœ์„ธ์Šค๋ฅผ ์‹คํ–‰ํ•˜๊ฑฐ๋‚˜ ๋ฌธ์ œ ํ•ด๊ฒฐ์„ ์œ„ํ•œ ์œ ํšจํ•œ ์ถ”๋ก  ๊ฐœ๋…์„ ๊ณ ์•ˆํ•  ์ˆ˜ ์žˆ๋Š” ๋Šฅ๋ ฅ์„ ํ‰๊ฐ€ํ•ฉ๋‹ˆ๋‹ค.
Next, we summarize the discussions of this survey, and introduce the challenges and future directions for LLMs, in the following aspects.
๋‹ค์Œ์œผ๋กœ, ๋ณธ ์„ค๋ฌธ์กฐ์‚ฌ์˜ ๋…ผ์˜๋ฅผ ์š”์•ฝํ•˜๊ณ , ์•ž์œผ๋กœ LLMs์ด ์ง๋ฉดํ•œ ๊ณผ์ œ์™€ ๋ฐœ์ „ ๋ฐฉํ–ฅ์„ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์ธก๋ฉด์—์„œ ์†Œ๊ฐœํ•œ๋‹ค.
๋‹ค์Œ์œผ๋กœ, ๋ณธ ์„ค๋ฌธ์กฐ์‚ฌ์˜ ๋…ผ์˜๋ฅผ ์š”์•ฝํ•˜๊ณ , ์•ž์œผ๋กœ LLM์ด ์ง๋ฉดํ•œ ๊ณผ์ œ์™€ ๋ฐœ์ „ ๋ฐฉํ–ฅ์„ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์ธก๋ฉด์—์„œ ์†Œ๊ฐœํ•œ๋‹ค.
First, let's determine the number of cherry trees Maria has. We are told she has 4 less than a third as many cherry trees as apple trees.
๋จผ์ € ๋งˆ๋ฆฌ์•„๊ฐ€ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๋ฒš๋‚˜๋ฌด์˜ ์ˆ˜๋ฅผ ์•Œ์•„๋ด…์‹œ๋‹ค. ๋งˆ๋ฆฌ์•„๋Š” ์‚ฌ๊ณผ๋‚˜๋ฌด์˜ 3๋ถ„์˜ 1๋ณด๋‹ค 4๊ทธ๋ฃจ ์ ์€ ๋ฒš๋‚˜๋ฌด๋ฅผ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
๋จผ์ € ๋งˆ๋ฆฌ์•„๊ฐ€ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๋ฒš๋‚˜๋ฌด์˜ ์ˆ˜๋ฅผ ์•Œ์•„๋ด…์‹œ๋‹ค. ๋งˆ๋ฆฌ์•„๋Š” ์‚ฌ๊ณผ๋‚˜๋ฌด๋ณด๋‹ค 3๋ถ„์˜ 1๋ณด๋‹ค ์ ์€ 4๊ทธ๋ฃจ์˜ ๋ฒš๋‚˜๋ฌด๋ฅผ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
An integer sns is odd if sn = 2k+1$ for some integer $k$.n\nTo prove that $A = B$, we must show that $A \subseteq B$ and $B \subseteq A$.
์ •์ˆ˜ sns๋Š” ์ผ๋ถ€ ์ •์ˆ˜ $k$์— ๋Œ€ํ•ด sn = 2k+1$์ด๋ฉด ํ™€์ˆ˜์ž…๋‹ˆ๋‹ค.nA = B$์ž„์„ ์ฆ๋ช…ํ•˜๋ ค๋ฉด $A \subseteq B$์™€ $B \subseteq A$๋ฅผ ๋ณด์—ฌ์ค˜์•ผ ํ•ฉ๋‹ˆ๋‹ค.
์ •์ˆ˜ sns๋Š” ์ผ๋ถ€ ์ •์ˆ˜ $k$์— ๋Œ€ํ•ด sn = 2k+1$์ด๋ฉด ํ™€์ˆ˜์ž…๋‹ˆ๋‹ค.nA = B$์ž„์„ ์ฆ๋ช…ํ•˜๋ ค๋ฉด $A \subseteq B$์™€ $B \subseteq A$๋ฅผ ๋ณด์—ฌ์ค˜์•ผ ํ•ฉ๋‹ˆ๋‹ค.
- Participating in the debate would give Trump a platform to outline his policies and vision, which could be beneficial for his campaign.
- ํ† ๋ก ์— ์ฐธ์—ฌํ•˜๋ฉด ํŠธ๋Ÿผํ”„๊ฐ€ ์ž์‹ ์˜ ์ •์ฑ…๊ณผ ๋น„์ „์„ ์„ค๋ช…ํ•  ์ˆ˜ ์žˆ๋Š” ํ”Œ๋žซํผ์„ ํ™•๋ณดํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋Š” ๊ทธ์˜ ์บ ํŽ˜์ธ์— ๋„์›€์ด ๋  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ํ† ๋ก ์— ์ฐธ์—ฌํ•˜๋ฉด ํŠธ๋Ÿผํ”„๊ฐ€ ์ž์‹ ์˜ ์ •์ฑ…๊ณผ ๋น„์ „์„ ์„ค๋ช…ํ•  ์ˆ˜ ์žˆ๋Š” ํ”Œ๋žซํผ์„ ํ™•๋ณดํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋Š” ๊ทธ์˜ ์บ ํŽ˜์ธ์— ๋„์›€์ด ๋  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Incoherent steps error (ISE). Incoherent chain of thoughts. Some steps in the generated chain of thought did not follow from prior ones.
๋ถˆ์ผ์น˜ ๋‹จ๊ณ„ ์˜ค๋ฅ˜(ISE). ์ผ๊ด€์„ฑ ์—†๋Š” ์ƒ๊ฐ์˜ ์‚ฌ์Šฌ. ์ƒ์„ฑ๋œ ์ƒ๊ฐ์˜ ์‚ฌ์Šฌ ์ค‘ ์ผ๋ถ€ ๋‹จ๊ณ„๊ฐ€ ์ด์ „ ๋‹จ๊ณ„๋ฅผ ๋”ฐ๋ฅด์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.
๋ถˆ์ผ์น˜ ๋‹จ๊ณ„ ์˜ค๋ฅ˜(ISE). ์ผ๊ด€์„ฑ ์—†๋Š” ์ƒ๊ฐ์˜ ์‚ฌ์Šฌ. ์ƒ์„ฑ๋œ ์ƒ๊ฐ์˜ ์‚ฌ์Šฌ ์ค‘ ์ผ๋ถ€ ๋‹จ๊ณ„๊ฐ€ ์ด์ „ ๋‹จ๊ณ„๋ฅผ ๋”ฐ๋ฅด์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.
Figure 10: **Reverse accuracy for Llama-1 models.** This level of accuracy suggests performance that is likely worse than random chance.
๊ทธ๋ฆผ 10: **Llama-1 ๋ชจ๋ธ์˜ ์—ญ ์ •ํ™•๋„.** ์ด ์ˆ˜์ค€์˜ ์ •ํ™•๋„๋Š” ๋ฌด์ž‘์œ„ ํ™•๋ฅ ๋ณด๋‹ค ๋‚˜์  ์ˆ˜ ์žˆ๋Š” ์„ฑ๋Šฅ์„ ์‹œ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 10: **Llama-1 ๋ชจ๋ธ์˜ ์—ญ ์ •ํ™•๋„.** ์ด ์ˆ˜์ค€์˜ ์ •ํ™•๋„๋Š” ๋ฌด์ž‘์œ„ ํ™•๋ฅ ๋ณด๋‹ค ๋‚˜์  ์ˆ˜ ์žˆ๋Š” ์„ฑ๋Šฅ์„ ์‹œ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
Where \(q\) is the question of a math problem. \(w_{1:T}\) denotes the sampled answer with a length of \(T\). \(p\) is the policy model.
์—ฌ๊ธฐ์„œ \(q\)๋Š” ์ˆ˜ํ•™ ๋ฌธ์ œ์˜ ์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค. \(w_{1:T}\)๋Š” ๊ธธ์ด๊ฐ€ \(T\)์ธ ์ƒ˜ํ”Œ๋ง๋œ ๋‹ต์„ ๋‚˜ํƒ€๋ƒ…๋‹ˆ๋‹ค. \(p\)๋Š” policy ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค.
์—ฌ๊ธฐ์„œ \(q\)๋Š” ์ˆ˜ํ•™ ๋ฌธ์ œ์˜ ์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค. w_{1:T}\)๋Š” ๊ธธ์ด๊ฐ€ \(T\)์ธ ์ƒ˜ํ”Œ๋ง๋œ ๋‹ต์„ ๋‚˜ํƒ€๋ƒ…๋‹ˆ๋‹ค. p\)๋Š” ์ •์ฑ… ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค.
Footnote 6: Note that TowerInstruct has used WMTโ€™22 test data for training, so we exclude it from comparison on the WMTโ€™22 test dataset.
์ฃผ 6: TowerInstruct๋Š” WMT'22 ํ…Œ์ŠคํŠธ ๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์— ์‚ฌ์šฉํ–ˆ์œผ๋ฏ€๋กœ WMT'22 ํ…Œ์ŠคํŠธ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ๋น„๊ต์—์„œ ์ œ์™ธํ•ฉ๋‹ˆ๋‹ค.
์ฃผ 6: TowerInstruct๋Š” WMT'22 ํ…Œ์ŠคํŠธ ๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์— ์‚ฌ์šฉํ–ˆ์œผ๋ฏ€๋กœ WMT'22 ํ…Œ์ŠคํŠธ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ๋น„๊ต์—์„œ ์ œ์™ธํ•ฉ๋‹ˆ๋‹ค.
Benefits of Mixing with AI Data.The above machinery allows us to analyze a particular regime where AI-data can help improve performance.
AI ๋ฐ์ดํ„ฐ์™€์˜ ํ˜ผํ•ฉ์˜ ์ด์ . ์œ„์˜ ๊ธฐ๊ณ„๋ฅผ ํ†ตํ•ด AI ๋ฐ์ดํ„ฐ๊ฐ€ ์„ฑ๋Šฅ ํ–ฅ์ƒ์— ๋„์›€์ด ๋  ์ˆ˜ ์žˆ๋Š” ํŠน์ • ์ฒด์ œ๋ฅผ ๋ถ„์„ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
AI ๋ฐ์ดํ„ฐ์™€์˜ ํ˜ผํ•ฉ์˜ ์ด์ .์œ„์˜ ๊ธฐ๊ณ„๋ฅผ ํ†ตํ•ด AI ๋ฐ์ดํ„ฐ๊ฐ€ ์„ฑ๋Šฅ ํ–ฅ์ƒ์— ๋„์›€์ด ๋  ์ˆ˜ ์žˆ๋Š” ํŠน์ • ์ฒด์ œ๋ฅผ ๋ถ„์„ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- Janet's ducks lay 16 eggs per day. - She eats three for breakfast every morning and bakes muffins for her friends every day with four.
- Janet์˜ ์˜ค๋ฆฌ๋Š” ํ•˜๋ฃจ์— 16๊ฐœ์˜ ์•Œ์„ ๋‚ณ์Šต๋‹ˆ๋‹ค. - ๊ทธ๋…€๋Š” ๋งค์ผ ์•„์นจ ์„ธ ๊ฐœ๋ฅผ ๋จน๊ณ  ์นœ๊ตฌ๋“ค์„ ์œ„ํ•ด ๋งค์ผ ๋„ค ๊ฐœ๋กœ ๋จธํ•€์„ ๊ตฝ์Šต๋‹ˆ๋‹ค.
- ์ž๋„ท์˜ ์˜ค๋ฆฌ๋Š” ํ•˜๋ฃจ์— 16๊ฐœ์˜ ์•Œ์„ ๋‚ณ์Šต๋‹ˆ๋‹ค. - ๊ทธ๋…€๋Š” ๋งค์ผ ์•„์นจ ์„ธ ๊ฐœ๋ฅผ ๋จน๊ณ  ์นœ๊ตฌ๋“ค์„ ์œ„ํ•ด ๋งค์ผ ๋„ค ๊ฐœ๋กœ ๋จธํ•€์„ ๊ตฝ์Šต๋‹ˆ๋‹ค.
Question: What was the revenue drop in the first quarter compared to the same period last year? Answer: The revenue dropped 15 percent.
์งˆ๋ฌธ: ์ž‘๋…„ ๊ฐ™์€ ๊ธฐ๊ฐ„๊ณผ ๋น„๊ตํ•˜์—ฌ 1๋ถ„๊ธฐ์— ์ˆ˜์ต์ด ์–ผ๋งˆ๋‚˜ ๊ฐ์†Œํ–ˆ๋‚˜์š”? ๋‹ต๋ณ€: ์ˆ˜์ต์ด 15% ๊ฐ์†Œํ–ˆ์Šต๋‹ˆ๋‹ค.
์งˆ๋ฌธ: ์ž‘๋…„ ๊ฐ™์€ ๊ธฐ๊ฐ„๊ณผ ๋น„๊ตํ•˜์—ฌ 1๋ถ„๊ธฐ์— ์ˆ˜์ต์ด ์–ผ๋งˆ๋‚˜ ๊ฐ์†Œํ–ˆ๋‚˜์š”? ๋‹ต๋ณ€: ์ˆ˜์ต์ด 15% ๊ฐ์†Œํ–ˆ์Šต๋‹ˆ๋‹ค.
I have this xml and i want to get the xml data to auto populate HTML table, the codes works but it makes duplicate on the table content
์ด xml์ด ์žˆ๊ณ  xml ๋ฐ์ดํ„ฐ๋ฅผ ๊ฐ€์ ธ์™€ HTML ํ…Œ์ด๋ธ”์„ ์ž๋™์œผ๋กœ ์ฑ„์šฐ๊ณ  ์‹ถ์€๋ฐ ์ฝ”๋“œ๊ฐ€ ์ž‘๋™ํ•˜์ง€๋งŒ ํ…Œ์ด๋ธ” ์ฝ˜ํ…์ธ ๊ฐ€ ์ค‘๋ณต๋ฉ๋‹ˆ๋‹ค.
์ด xml์ด ์žˆ๊ณ  xml ๋ฐ์ดํ„ฐ๋ฅผ ๊ฐ€์ ธ์™€ HTML ํ…Œ์ด๋ธ”์„ ์ž๋™์œผ๋กœ ์ฑ„์šฐ๊ณ  ์‹ถ์€๋ฐ ์ฝ”๋“œ๊ฐ€ ์ž‘๋™ํ•˜์ง€๋งŒ ํ…Œ์ด๋ธ” ์ฝ˜ํ…์ธ ๊ฐ€ ์ค‘๋ณต๋ฉ๋‹ˆ๋‹ค.
Some of the popular frameworks which are useful for LLM training includes (note that some of them can be used beyond LLM training too):
LLM ํ•™์Šต์— ์œ ์šฉํ•œ ์ธ๊ธฐ ์žˆ๋Š” ํ”„๋ ˆ์ž„์›Œํฌ์—๋Š” ๋‹ค์Œ์ด ํฌํ•จ๋ฉ๋‹ˆ๋‹ค(์ด ์ค‘ ์ผ๋ถ€๋Š” LLM ํ•™์Šต ์™ธ์—๋„ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค).
LLM ํ•™์Šต์— ์œ ์šฉํ•œ ์ธ๊ธฐ ์žˆ๋Š” ํ”„๋ ˆ์ž„์›Œํฌ์—๋Š” ๋‹ค์Œ์ด ํฌํ•จ๋ฉ๋‹ˆ๋‹ค(์ด ์ค‘ ์ผ๋ถ€๋Š” LLM ํ•™์Šต ์™ธ์—๋„ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค).
3. There are twice the number of emeralds than the rubies. We can find the number of emeralds by multiplying the number of rubies by 2.
3. ๋ฃจ๋น„๋ณด๋‹ค ์—๋ฉ”๋ž„๋“œ๊ฐ€ ๋‘ ๋ฐฐ ๋” ๋งŽ์Šต๋‹ˆ๋‹ค. ๋ฃจ๋น„์˜ ์ˆ˜์— 2๋ฅผ ๊ณฑํ•˜๋ฉด ์—๋ฉ”๋ž„๋“œ์˜ ์ˆ˜๋ฅผ ๊ตฌํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
3. ๋ฃจ๋น„๋ณด๋‹ค ์—๋ฉ”๋ž„๋“œ๊ฐ€ ๋‘ ๋ฐฐ ๋” ๋งŽ์Šต๋‹ˆ๋‹ค. ๋ฃจ๋น„์˜ ์ˆ˜์— 2๋ฅผ ๊ณฑํ•˜๋ฉด ์—๋ฉ”๋ž„๋“œ์˜ ์ˆ˜๋ฅผ ๊ตฌํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
where \(\left(\frac{N_{c}}{N}\right)^{\alpha_{N}}\) is Eq. 3.1, and \(S_{c}\) and \(\alpha_{S}\) are constant scalars to be estimated.
์—ฌ๊ธฐ์„œ \(\left(\frac{N_{c}}{N}\right)^{\alpha_{N}}\)๋Š” ์‹ 3.1์ด๊ณ , \(S_{c}\)์™€ \(\alpha_{S}\)๋Š” ์ถ”์ •ํ•ด์•ผ ํ•  ์ƒ์ˆ˜ ์Šค์นผ๋ผ์ด๋‹ค.
์—ฌ๊ธฐ์„œ \(\left(\frac{N_{c}}{N}\right)^{\alpha_{N}}\)๋Š” ์‹ 3.1์ด๊ณ , \(S_{c}\)์™€ \(\alpha_{S}\)๋Š” ์ถ”์ •ํ•ด์•ผ ํ•  ์ƒ์ˆ˜ ์Šค์นผ๋ผ์ด๋‹ค.
_Synthetic QAs: question 1:_ "Who was John Mayne?" _answer 1:_ "John Mayne was a Scottish printer, journalist, and poet from Dumfries.
_ํ•ฉ์„ฑ QAs: ์งˆ๋ฌธ 1:_ "์กด ๋ฉ”์ธ์€ ๋ˆ„๊ตฌ์˜€๋‚˜์š”?" _๋‹ต๋ณ€ 1:_ "์กด ๋ฉ”์ธ์€ ๋คํ”„๋ฆฌ์Šค ์ถœ์‹ ์˜ ์Šค์ฝ”ํ‹€๋žœ๋“œ ์ธ์‡„์—…์ž, ์–ธ๋ก ์ธ, ์‹œ์ธ์ด์—ˆ์Šต๋‹ˆ๋‹ค.
_ํ•ฉ์„ฑ QA: ์งˆ๋ฌธ 1:_ "์กด ๋ฉ”์ธ์€ ๋ˆ„๊ตฌ์˜€๋‚˜์š”?" _๋‹ต๋ณ€ 1:_ "์กด ๋ฉ”์ธ์€ ๋คํ”„๋ฆฌ์Šค ์ถœ์‹ ์˜ ์Šค์ฝ”ํ‹€๋žœ๋“œ ์ธ์‡„์—…์ž, ์–ธ๋ก ์ธ, ์‹œ์ธ์ด์—ˆ์Šต๋‹ˆ๋‹ค.
Yekyung Kim\({}^{\circledS}\), Yapei Chang\({}^{\circledS}\), Marzena Karpinska\({}^{\circledS}\), Aparna Garimella\({}^{\circledS}\),
๊น€์˜ˆ๊ฒฝ\({}^{\circledS}\), ์•ผํŽ˜์ด ์ฐฝ\({}^{\circledS}\), ๋งˆ๋ฅด์ œ๋‚˜ ์นด๋ฅดํ•€์Šค์นด\({}^{\circledS}\), ์•„ํŒŒ๋ฅด๋‚˜ ๊ฐ€๋ฆฌ๋ฉœ๋ผ\({}^{\circledS}\),
๊น€์˜ˆ๊ฒฝ\({}^{\circledS}\), ์•ผํŽ˜์ด ์ฐฝ\({}^{\circledS}\), ๋งˆ๋ฅด์ œ๋‚˜ ์นด๋ฅดํ•€์Šค์นด\({}^{\circledS}\), ์•„ํŒŒ๋ฅด๋‚˜ ๊ฐ€๋ฆฌ๋ฉœ๋ผ\({}^{\circledS}\),
We say that the model \(\mathcal{M}\) possesses _any_ knowledge regarding the set of questions \(\mathcal{Q}\) if the following holds:
๋‹ค์Œ์ด ์„ฑ๋ฆฝํ•˜๋Š” ๊ฒฝ์šฐ, ์šฐ๋ฆฌ๋Š” ๋ชจ๋ธ \(\mathcal{M}\)์ด ์งˆ๋ฌธ ์ง‘ํ•ฉ \(\mathcal{Q}\)์— ๊ด€ํ•œ _any_ ์ง€์‹์„ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค๊ณ  ๋งํ•ฉ๋‹ˆ๋‹ค:
๋‹ค์Œ์ด ์„ฑ๋ฆฝํ•˜๋Š” ๊ฒฝ์šฐ, ์šฐ๋ฆฌ๋Š” ๋ชจ๋ธ \(mathcal{M}\)์ด ์งˆ๋ฌธ ์ง‘ํ•ฉ \(mathcal{Q}\)์— ๊ด€ํ•œ _๋ชจ๋“ _ ์ง€์‹์„ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค๊ณ  ๋งํ•ฉ๋‹ˆ๋‹ค:
- International diplomatic efforts or significant pressure on Israel to de-escalate could decrease the probability of a 'Yes' outcome.
- ๊ตญ์ œ ์™ธ๊ต์  ๋…ธ๋ ฅ์ด๋‚˜ ์ด์Šค๋ผ์—˜์ด ํ™•์ „์„ ์–ต์ œํ•˜๋„๋ก ์ƒ๋‹นํ•œ ์••๋ ฅ์„ ๊ฐ€ํ•˜๋Š” ๊ฒƒ์€ '๊ทธ๋ ‡๋‹ค' ๊ฒฐ๊ณผ์˜ ํ™•๋ฅ ์„ ๋‚ฎ์ถœ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ๊ตญ์ œ ์™ธ๊ต์  ๋…ธ๋ ฅ์ด๋‚˜ ์ด์Šค๋ผ์—˜์ด ํ™•์ „์„ ์–ต์ œํ•˜๋„๋ก ์ƒ๋‹นํ•œ ์••๋ ฅ์„ ๊ฐ€ํ•˜๋Š” ๊ฒƒ์€ '๊ทธ๋ ‡๋‹ค' ๊ฒฐ๊ณผ์˜ ํ™•๋ฅ ์„ ๋‚ฎ์ถœ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
translation scores reflect genuinely better translations or if they simply align more closely with the evaluation model's preferences.
๋ฒˆ์—ญ ์ ์ˆ˜๊ฐ€ ์‹ค์ œ๋กœ ๋” ๋‚˜์€ ๋ฒˆ์—ญ์„ ๋ฐ˜์˜ํ•˜๋Š”์ง€ ์•„๋‹ˆ๋ฉด ํ‰๊ฐ€ ๋ชจ๋ธ์˜ ์„ ํ˜ธ๋„์— ๋” ๊ฐ€๊น๊ฒŒ ๋ถ€ํ•ฉํ•˜๋Š”์ง€ ์—ฌ๋ถ€๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค.
๋ฒˆ์—ญ ์ ์ˆ˜๊ฐ€ ์‹ค์ œ๋กœ ๋” ๋‚˜์€ ๋ฒˆ์—ญ์„ ๋ฐ˜์˜ํ•˜๋Š”์ง€ ์•„๋‹ˆ๋ฉด ํ‰๊ฐ€ ๋ชจ๋ธ์˜ ์„ ํ˜ธ๋„์— ๋” ๊ฐ€๊น๊ฒŒ ๋ถ€ํ•ฉํ•˜๋Š”์ง€ ์—ฌ๋ถ€๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค.
In this part, we review some of the popular approaches used for more cost-friendly (and compute-friendly) training and usage of LLMs.
์ด ๋ถ€๋ถ„์—์„œ๋Š” LLM์„ ๋ณด๋‹ค ๋น„์šฉ ์นœํ™”์ ์ด๊ณ  ์ปดํ“จํŒ… ์นœํ™”์ ์œผ๋กœ ํ›ˆ๋ จํ•˜๊ณ  ์‚ฌ์šฉํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋˜๋Š” ๋ช‡ ๊ฐ€์ง€ ์ธ๊ธฐ ์žˆ๋Š” ์ ‘๊ทผ ๋ฐฉ์‹์„ ๊ฒ€ํ† ํ•ฉ๋‹ˆ๋‹ค.
์ด ๋ถ€๋ถ„์—์„œ๋Š” LLM์„ ๋ณด๋‹ค ๋น„์šฉ ์นœํ™”์ ์ด๊ณ  ์ปดํ“จํŒ… ์นœํ™”์ ์œผ๋กœ ํ›ˆ๋ จํ•˜๊ณ  ์‚ฌ์šฉํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋˜๋Š” ๋ช‡ ๊ฐ€์ง€ ์ธ๊ธฐ ์žˆ๋Š” ์ ‘๊ทผ ๋ฐฉ์‹์„ ๊ฒ€ํ† ํ•ฉ๋‹ˆ๋‹ค.
Figure 15 \(|\) R-GSM example where both the original and the reordered problems were correctly solved by all LLMs in our evaluation.
๊ทธ๋ฆผ 15 \(|\) R-GSM ์˜ˆ์ œ์—์„œ ์›๋ž˜ ๋ฌธ์ œ์™€ ์žฌ์ •๋ ฌ๋œ ๋ฌธ์ œ๋ฅผ ๋ชจ๋‘ ํ‰๊ฐ€์— ์ฐธ์—ฌํ•œ ๋ชจ๋“  LLM์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ํ’€์—ˆ์Šต๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 15 \(|\) R-GSM ์˜ˆ์ œ์—์„œ ์›๋ž˜ ๋ฌธ์ œ์™€ ์žฌ์ •๋ ฌ๋œ ๋ฌธ์ œ๋ฅผ ๋ชจ๋‘ ํ‰๊ฐ€์— ์ฐธ์—ฌํ•œ ๋ชจ๋“  LLM์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ํ’€์—ˆ์Šต๋‹ˆ๋‹ค.
Figure 6: Illustration of Result 8 (see Figure 14 for \(\mathsf{bioS}\) data and Figure 15 for \(\mathsf{bioD}\) data in Appendix C).
๊ทธ๋ฆผ 6: ๊ฒฐ๊ณผ 8์˜ ์˜ˆ์‹œ(๋ถ€๋ก C์˜ ๊ทธ๋ฆผ 14๋Š” \(\mathsf{bioS}\) ๋ฐ์ดํ„ฐ, ๊ทธ๋ฆผ 15๋Š” \(\mathsf{bioD}\) ๋ฐ์ดํ„ฐ ์ฐธ์กฐ).
๊ทธ๋ฆผ 6: ๊ฒฐ๊ณผ 8์˜ ์˜ˆ์‹œ(๋ถ€๋ก C์˜ ๊ทธ๋ฆผ 14๋Š” \(\mathsf{bioS}\) ๋ฐ์ดํ„ฐ, ๊ทธ๋ฆผ 15๋Š” \(\mathsf{bioD}\) ๋ฐ์ดํ„ฐ ์ฐธ์กฐ).
Proof.: Let \(e_{0j_{1}},e_{j_{1}j_{2}},\ldots,e_{j_{m-1}i}\) be the edges of the minimum distance path from \(0\) to \(i\) in \(G\).
์ฆ๋ช….: \(e_{0j_{1}},e_{j_{1}j_{2}},\ldots,e_{j_{m-1}i}\)๋ฅผ \(G\)์—์„œ \(0\)์—์„œ \(i\)๊นŒ์ง€์˜ ์ตœ์†Œ ๊ฑฐ๋ฆฌ ๊ฒฝ๋กœ์˜ ๊ฐ€์žฅ์ž๋ฆฌ๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
์ฆ๋ช….: e_{0j_{1}},e_{j_{1}j_{2}},\ldots,e_{j_{m-1}i}\)๋ฅผ \(G\)์—์„œ \(0\)์—์„œ \(i\)๊นŒ์ง€์˜ ์ตœ์†Œ ๊ฑฐ๋ฆฌ ๊ฒฝ๋กœ์˜ ๊ฐ€์žฅ์ž๋ฆฌ๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
Figure 14 \(|\) R-GSM example where both the original and the reordered problems were correctly solved by all LLMs in our evaluation.
๊ทธ๋ฆผ 14 \(|\) R-GSM ์˜ˆ์ œ์—์„œ ์›๋ž˜ ๋ฌธ์ œ์™€ ์žฌ์ •๋ ฌ๋œ ๋ฌธ์ œ๋ฅผ ๋ชจ๋‘ ํ‰๊ฐ€์— ์ฐธ์—ฌํ•œ ๋ชจ๋“  LLM์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ํ’€์—ˆ์Šต๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 14 \(|\) R-GSM ์˜ˆ์ œ์—์„œ ์›๋ž˜ ๋ฌธ์ œ์™€ ์žฌ์ •๋ ฌ๋œ ๋ฌธ์ œ๋ฅผ ๋ชจ๋‘ ํ‰๊ฐ€์— ์ฐธ์—ฌํ•œ ๋ชจ๋“  LLM์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ํ’€์—ˆ์Šต๋‹ˆ๋‹ค.
Project leader. Chen, Weiqi, Jingcheng and Yixuan are interns at MSRA. GitHub: Xwin-Math This repository will be continually updated.
ํ”„๋กœ์ ํŠธ ๋ฆฌ๋”. Chen, Weiqi, Jingcheng ๋ฐ Yixuan์€ MSRA์˜ ์ธํ„ด์ž…๋‹ˆ๋‹ค. GitHub: Xwin-Math ์ด ๋ฆฌํฌ์ง€ํ† ๋ฆฌ๋Š” ์ง€์†์ ์œผ๋กœ ์—…๋ฐ์ดํŠธ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
ํ”„๋กœ์ ํŠธ ๋ฆฌ๋”. Chen, Weiqi, Jingcheng ๋ฐ Yixuan์€ MSRA์˜ ์ธํ„ด์ž…๋‹ˆ๋‹ค. GitHub: Xwin-Math ์ด ๋ฆฌํฌ์ง€ํ† ๋ฆฌ๋Š” ์ง€์†์ ์œผ๋กœ ์—…๋ฐ์ดํŠธ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
Figure 5: Stress-testing the LayerNorm extraction attack on models behind an API (a-b), and models using both RMSNorm and biases (c).
๊ทธ๋ฆผ 5: API ๋’ค์— ์žˆ๋Š” ๋ชจ๋ธ(a-b)๊ณผ RMSNorm๊ณผ ๋ฐ”์ด์–ด์Šค๋ฅผ ๋ชจ๋‘ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ๋ธ(c)์— ๋Œ€ํ•œ LayerNorm ์ถ”์ถœ ๊ณต๊ฒฉ์„ ์ŠคํŠธ๋ ˆ์Šค ํ…Œ์ŠคํŠธํ•ฉ๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 5: API ๋’ค์— ์žˆ๋Š” ๋ชจ๋ธ(a-b)๊ณผ RMSNorm๊ณผ ๋ฐ”์ด์–ด์Šค๋ฅผ ๋ชจ๋‘ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ๋ธ(c)์— ๋Œ€ํ•œ LayerNorm ์ถ”์ถœ ๊ณต๊ฒฉ์„ ์ŠคํŠธ๋ ˆ์Šค ํ…Œ์ŠคํŠธํ•ฉ๋‹ˆ๋‹ค.
Given a fixed model, the computational costs depend on the vocabulary size and the fertility of the tokenizer, as defined in Eq. (1).
๊ณ ์ •๋œ ๋ชจ๋ธ์ด ์ฃผ์–ด์ง€๋ฉด, ๊ณ„์‚ฐ ๋น„์šฉ์€ (1) ์‹์— ์ •์˜๋œ ๋ฐ”์™€ ๊ฐ™์ด ์–ดํœ˜ ํฌ๊ธฐ์™€ tokenizer์˜ ์ƒ์‹๋ ฅ์— ๋”ฐ๋ผ ๋‹ฌ๋ผ์ง„๋‹ค.
๊ณ ์ •๋œ ๋ชจ๋ธ์ด ์ฃผ์–ด์ง€๋ฉด, ๊ณ„์‚ฐ ๋น„์šฉ์€ (1) ์‹์— ์ •์˜๋œ ๋ฐ”์™€ ๊ฐ™์ด ์–ดํœ˜ ํฌ๊ธฐ์™€ ํ† ํฐํ™”๊ธฐ์˜ ์ƒ์‹๋ ฅ์— ๋”ฐ๋ผ ๋‹ฌ๋ผ์ง„๋‹ค.
In this part, we first formulate the general planning paradigm of LLMs for solving complex tasks, which is illustrated in Figure 16.
์ด ๋ถ€๋ถ„์—์„œ๋Š” ๋จผ์ € ๋ณต์žกํ•œ ๊ณผ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ LLMs์˜ ์ผ๋ฐ˜์ ์ธ ๊ณ„ํš ํŒจ๋Ÿฌ๋‹ค์ž„์„ ๊ณต์‹ํ™”ํ•˜๋Š”๋ฐ, ์ด๋Š” ๊ทธ๋ฆผ 16์— ์„ค๋ช…๋˜์–ด ์žˆ๋‹ค.
์ด ๋ถ€๋ถ„์—์„œ๋Š” ๋จผ์ € ๋ณต์žกํ•œ ๊ณผ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ LLM์˜ ์ผ๋ฐ˜์ ์ธ ๊ณ„ํš ํŒจ๋Ÿฌ๋‹ค์ž„์„ ๊ณต์‹ํ™”ํ•˜๋Š”๋ฐ, ์ด๋Š” ๊ทธ๋ฆผ 16์— ์„ค๋ช…๋˜์–ด ์žˆ๋‹ค.
- She has 4 less than a third as many cherry trees as apple trees. A third of the apple trees is 80/3, which is approximately 26.67.
- ์‚ฌ๊ณผ๋‚˜๋ฌด๋ณด๋‹ค ์ฒด๋ฆฌ๋‚˜๋ฌด๊ฐ€ 3๋ถ„์˜ 1๋„ ์•ˆ ๋˜๋Š” 4๊ทธ๋ฃจ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. ์‚ฌ๊ณผ๋‚˜๋ฌด์˜ 3๋ถ„์˜ 1์€ 80/3์œผ๋กœ ์•ฝ 26.67์ž…๋‹ˆ๋‹ค.
- ์‚ฌ๊ณผ๋‚˜๋ฌด๋ณด๋‹ค ์ฒด๋ฆฌ๋‚˜๋ฌด๊ฐ€ 3๋ถ„์˜ 1๋„ ์•ˆ ๋˜๋Š” 4๊ทธ๋ฃจ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. ์‚ฌ๊ณผ๋‚˜๋ฌด์˜ 3๋ถ„์˜ 1์€ 80/3์œผ๋กœ ์•ฝ 26.67์ž…๋‹ˆ๋‹ค.
resolution (Figure 3(c)). This may be due to the aforementioned issue: Our model hedges, even as the evidence becomes more decisive.
ํ•ด์ƒ๋„(๊ทธ๋ฆผ 3(c))๋ฅผ ๋ณด์—ฌ์ค€๋‹ค. ์ด๋Š” ์•ž์„œ ์–ธ๊ธ‰ํ•œ ๋ฌธ์ œ ๋•Œ๋ฌธ์ผ ์ˆ˜ ์žˆ๋‹ค: ์šฐ๋ฆฌ ๋ชจ๋ธ์€ ์ฆ๊ฑฐ๊ฐ€ ๋” ๊ฒฐ์ •์ ์ด ๋˜๋”๋ผ๋„ hedgesํ•œ๋‹ค.
ํ•ด์ƒ๋„(๊ทธ๋ฆผ 3(c))๋ฅผ ๋ณด์—ฌ์ค€๋‹ค. ์ด๋Š” ์•ž์„œ ์–ธ๊ธ‰ํ•œ ๋ฌธ์ œ ๋•Œ๋ฌธ์ผ ์ˆ˜ ์žˆ๋‹ค: ์šฐ๋ฆฌ ๋ชจ๋ธ์€ ์ฆ๊ฑฐ๊ฐ€ ๋” ๊ฒฐ์ •์ ์ด ๋˜๋”๋ผ๋„ ํ—ค์ง€ํ•œ๋‹ค.
and we recover the classical Hutter result! Thus, our test metric defined in (17) is pointing in the right direction, conceptually.
๊ทธ๋ฆฌ๊ณ  ์šฐ๋ฆฌ๋Š” ๊ณ ์ „์ ์ธ Hutter์˜ ๊ฒฐ๊ณผ๋ฅผ ํšŒ๋ณตํ•œ๋‹ค! ๋”ฐ๋ผ์„œ (17)์—์„œ ์ •์˜ํ•œ ํ…Œ์ŠคํŠธ ๋ฉ”ํŠธ๋ฆญ์€ ๊ฐœ๋…์ ์œผ๋กœ ์˜ฌ๋ฐ”๋ฅธ ๋ฐฉํ–ฅ์„ ๊ฐ€๋ฆฌํ‚ค๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
๊ทธ๋ฆฌ๊ณ  ์šฐ๋ฆฌ๋Š” ๊ณ ์ „์ ์ธ ํ—ˆํ„ฐ์˜ ๊ฒฐ๊ณผ๋ฅผ ํšŒ๋ณตํ•œ๋‹ค! ๋”ฐ๋ผ์„œ (17)์—์„œ ์ •์˜ํ•œ ํ…Œ์ŠคํŠธ ๋ฉ”ํŠธ๋ฆญ์€ ๊ฐœ๋…์ ์œผ๋กœ ์˜ฌ๋ฐ”๋ฅธ ๋ฐฉํ–ฅ์„ ๊ฐ€๋ฆฌํ‚ค๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
Footnote 1: All resources are publicly available at [https://github.com/tLabruna/Adapt-LLM](https://github.com/tLabruna/Adapt-LLM).
๊ฐ์ฃผ 1: ๋ชจ๋“  ๋ฆฌ์†Œ์Šค๋Š” [https://github.com/tLabruna/Adapt-LLM](https://github.com/tLabruna/Adapt-LLM)์—์„œ ๊ณต๊ฐœ์ ์œผ๋กœ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
๊ฐ์ฃผ 1: ๋ชจ๋“  ๋ฆฌ์†Œ์Šค๋Š” [https://github.com/tLabruna/Adapt-LLM](https://github.com/tLabruna/Adapt-LLM)์—์„œ ๊ณต๊ฐœ์ ์œผ๋กœ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Fig. 27: Subsequent stages of Macrodata Refinement remove nearly 90% of the documents originally in CommonCrawl. Courtesy of [124].
๊ทธ๋ฆผ 27: ๋งคํฌ๋กœ๋ฐ์ดํ„ฐ ๊ตฌ์ฒดํ™”์˜ ํ›„์† ๋‹จ๊ณ„์—์„œ๋Š” ์›๋ž˜ CommonCrawl์— ์žˆ๋˜ ๋ฌธ์„œ์˜ ๊ฑฐ์˜ 90%๊ฐ€ ์ œ๊ฑฐ๋ฉ๋‹ˆ๋‹ค. [124]์˜ ๋•๋ถ„์ž…๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 27: ๋งคํฌ๋กœ๋ฐ์ดํ„ฐ ๊ตฌ์ฒดํ™”์˜ ํ›„์† ๋‹จ๊ณ„์—์„œ๋Š” ์›๋ž˜ CommonCrawl์— ์žˆ๋˜ ๋ฌธ์„œ์˜ ๊ฑฐ์˜ 90%๊ฐ€ ์ œ๊ฑฐ๋ฉ๋‹ˆ๋‹ค. [124]์˜ ๋•๋ถ„์ž…๋‹ˆ๋‹ค.
Lower-Bound.WLOG6 consider the following specific choice of permutations defined by \(\pi_{i}(j)=j\) (i.e doesn't depend on \(i\)).
Lower-Bound.WLOG6๋Š” \(\pi_{i}(j)=j\)(์ฆ‰, \(i\)์— ์˜์กดํ•˜์ง€ ์•Š์Œ)๋กœ ์ •์˜๋œ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ํŠน์ • ์ˆœ์—ด์˜ ์„ ํƒ์„ ๊ณ ๋ คํ•ฉ๋‹ˆ๋‹ค.
Lower-Bound.WLOG6๋Š” \(\pi_{i}(j)=j\)(์ฆ‰, \(i\)์— ์˜์กดํ•˜์ง€ ์•Š์Œ)๋กœ ์ •์˜๋œ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ํŠน์ • ์ˆœ์—ด์˜ ์„ ํƒ์„ ๊ณ ๋ คํ•ฉ๋‹ˆ๋‹ค.
**Question:** How did Matthew Belk, a National Weather Service meteorologist, describe the September 2023 northeastern U.S. floods?
**Question:** ๋ฏธ๊ตญ ๊ตญ๋ฆฝ๊ธฐ์ƒ์ฒญ ๊ธฐ์ƒํ•™์ž์ธ ๋งคํŠœ ๋ฒจํฌ๋Š” 2023๋…„ 9์›” ๋ฏธ๊ตญ ๋ถ๋™๋ถ€ ํ™์ˆ˜๋ฅผ ์–ด๋–ป๊ฒŒ ์„ค๋ช…ํ–ˆ๋‚˜์š”?
**์งˆ๋ฌธ:** ๋ฏธ๊ตญ ๊ตญ๋ฆฝ๊ธฐ์ƒ์ฒญ ๊ธฐ์ƒํ•™์ž์ธ ๋งคํŠœ ๋ฒจํฌ๋Š” 2023๋…„ 9์›” ๋ฏธ๊ตญ ๋ถ๋™๋ถ€ ํ™์ˆ˜๋ฅผ ์–ด๋–ป๊ฒŒ ์„ค๋ช…ํ–ˆ๋‚˜์š”?
Jurning meets this report that a sheer of standing increases only inside white attack light, serving an acronym model in the story.
์ฃผ์–ด์ง„ ๋ณด๊ณ ์„œ์— ๋”ฐ๋ฅด๋ฉด, ์ˆœ์ „ํžˆ ์„œ ์žˆ๋Š” ์‚ฌ๋žŒ์€ ๋ฐฑ์ƒ‰ ๊ณต๊ฒฉ ์กฐ๋ช… ๋‚ด๋ถ€์—์„œ๋งŒ ์ฆ๊ฐ€ํ•˜๋ฉฐ, ์ด๋Š” ์Šคํ† ๋ฆฌ์—์„œ acronym ๋ชจ๋ธ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
์ฃผ์–ด์ง„ ๋ณด๊ณ ์„œ์— ๋”ฐ๋ฅด๋ฉด, ์ˆœ์ „ํžˆ ์„œ ์žˆ๋Š” ์‚ฌ๋žŒ์€ ๋ฐฑ์ƒ‰ ๊ณต๊ฒฉ ์กฐ๋ช… ๋‚ด๋ถ€์—์„œ๋งŒ ์ฆ๊ฐ€ํ•˜๋ฉฐ, ์ด๋Š” ์Šคํ† ๋ฆฌ์—์„œ ์•ฝ์–ด ๋ชจ๋ธ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
To understand our goal better, we expand on our analogy between training an LLM in the real-world setting of preparing for an exam.
๋ชฉํ‘œ๋ฅผ ๋” ์ž˜ ์ดํ•ดํ•˜๊ธฐ ์œ„ํ•ด, ์šฐ๋ฆฌ๋Š” ์‹œํ—˜์„ ์ค€๋น„ํ•˜๋Š” ์‹ค์ œ ํ™˜๊ฒฝ์—์„œ LLM์„ ํ›ˆ๋ จํ•˜๋Š” ๊ฒƒ๊ณผ์˜ ๋น„์œ ๋ฅผ ํ™•์žฅํ•œ๋‹ค.
๋ชฉํ‘œ๋ฅผ ๋” ์ž˜ ์ดํ•ดํ•˜๊ธฐ ์œ„ํ•ด, ์šฐ๋ฆฌ๋Š” ์‹œํ—˜์„ ์ค€๋น„ํ•˜๋Š” ์‹ค์ œ ํ™˜๊ฒฝ์—์„œ LLM์„ ํ›ˆ๋ จํ•˜๋Š” ๊ฒƒ๊ณผ์˜ ๋น„์œ ๋ฅผ ํ™•์žฅํ•œ๋‹ค.
**Question**: How close did the doctor's differential diagnosis (DDx) come to including the PROBABLE DIAGNOSIS from the answer key?
**์งˆ๋ฌธ**: ์˜์‚ฌ์˜ ๊ฐ๋ณ„ ์ง„๋‹จ(DDx)์ด ๋‹ต์•ˆ์ง€์˜ PROBABLE DIAGNOSIS์— ์–ผ๋งˆ๋‚˜ ๊ทผ์ ‘ํ–ˆ๋‚˜์š”?
**์งˆ๋ฌธ**: ์˜์‚ฌ์˜ ๊ฐ๋ณ„ ์ง„๋‹จ(DDx)์ด ๋‹ต์•ˆ์ง€์˜ PROBABLE DIAGNOSIS์— ์–ผ๋งˆ๋‚˜ ๊ทผ์ ‘ํ–ˆ๋‚˜์š”?
- Unforeseen events could negatively impact 'Barbie's attendance, such as controversies or widespread technical issues in theaters.
- ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ์‚ฌ๊ฑด(์˜ˆ: ๋…ผ๋ž€์ด๋‚˜ ๊ทน์žฅ์˜ ๊ด‘๋ฒ”์œ„ํ•œ ๊ธฐ์ˆ  ๋ฌธ์ œ ๋“ฑ)์ด ๋ฐœ์ƒํ•˜์—ฌ 'Barbie'์˜ ๊ด€๊ฐ ์ˆ˜์— ๋ถ€์ •์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ์‚ฌ๊ฑด(์˜ˆ: ๋…ผ๋ž€์ด๋‚˜ ๊ทน์žฅ์˜ ๊ด‘๋ฒ”์œ„ํ•œ ๊ธฐ์ˆ  ๋ฌธ์ œ ๋“ฑ)์ด ๋ฐœ์ƒํ•˜์—ฌ '๋ฐ”๋น„'์˜ ๊ด€๊ฐ ์ˆ˜์— ๋ถ€์ •์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Figure 8: Performance of Baichuan-7B, DeepSeek-7B, DeepSeek-67B, Amber-7B, OpenLLaMA-7B, and Yi-34B models across code benchmarks.
๊ทธ๋ฆผ 8: ์ฝ”๋“œ ๋ฒค์น˜๋งˆํฌ์—์„œ Baichuan-7B, DeepSeek-7B, DeepSeek-67B, Amber-7B, OpenLLaMA-7B ๋ฐ Yi-34B ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ.
๊ทธ๋ฆผ 8: ์ฝ”๋“œ ๋ฒค์น˜๋งˆํฌ์—์„œ Baichuan-7B, DeepSeek-7B, DeepSeek-67B, Amber-7B, OpenLLaMA-7B ๋ฐ Yi-34B ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ.
21. The story of โ€œDivine Rivalsโ€ explores themes of love, loyalty, and the difficult choices faced by individuals in times of war.
21. "์‹ ์„ฑํ•œ ๋ผ์ด๋ฒŒ"์˜ ์ด์•ผ๊ธฐ๋Š” ์‚ฌ๋ž‘, ์ถฉ์„ฑ์‹ฌ, ์ „์Ÿ ์ค‘ ๊ฐœ์ธ์ด ์ง๋ฉดํ•˜๋Š” ์–ด๋ ค์šด ์„ ํƒ์ด๋ผ๋Š” ์ฃผ์ œ๋ฅผ ํƒ๊ตฌํ•ฉ๋‹ˆ๋‹ค.
21. "์‹ ์„ฑํ•œ ๋ผ์ด๋ฒŒ"์˜ ์ด์•ผ๊ธฐ๋Š” ์‚ฌ๋ž‘, ์ถฉ์„ฑ์‹ฌ, ์ „์Ÿ ์ค‘ ๊ฐœ์ธ์ด ์ง๋ฉดํ•˜๋Š” ์–ด๋ ค์šด ์„ ํƒ์ด๋ผ๋Š” ์ฃผ์ œ๋ฅผ ํƒ๊ตฌํ•ฉ๋‹ˆ๋‹ค.
Figure 9: Performance of Baichuan-7B, DeepSeek-7B, DeepSeek-67B, Amber-7B, OpenLLaMA-7B, and Yi-34B models across math benchmarks.
๊ทธ๋ฆผ 9: ์ˆ˜ํ•™ ๋ฒค์น˜๋งˆํฌ์—์„œ Baichuan-7B, DeepSeek-7B, DeepSeek-67B, Amber-7B, OpenLLaMA-7B ๋ฐ Yi-34B ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ.
๊ทธ๋ฆผ 9: ์ˆ˜ํ•™ ๋ฒค์น˜๋งˆํฌ์—์„œ Baichuan-7B, DeepSeek-7B, DeepSeek-67B, Amber-7B, OpenLLaMA-7B ๋ฐ Yi-34B ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ.
We introduce a token selection ratio \(k\%\), which determines the proportion of tokens to be included based on their excess loss.
ํ† ํฐ ์„ ํƒ ๋น„์œจ \(k\%\)์„ ๋„์ž…ํ•˜์—ฌ ์ดˆ๊ณผ Loss๋ฅผ ๊ธฐ์ค€์œผ๋กœ ํฌํ•จํ•  ํ† ํฐ์˜ ๋น„์ค‘์„ ๊ฒฐ์ •ํ•œ๋‹ค.
ํ† ํฐ ์„ ํƒ ๋น„์œจ \(k\%\)์„ ๋„์ž…ํ•˜์—ฌ ์ดˆ๊ณผ ์†์‹ค์„ ๊ธฐ์ค€์œผ๋กœ ํฌํ•จํ•  ํ† ํฐ์˜ ๋น„์ค‘์„ ๊ฒฐ์ •ํ•œ๋‹ค.
21. The story of "Divine Rivals" explores themes of love, loyalty, and the difficult choices faced by individuals in times of war.
21. "์‹ ์„ฑํ•œ ๋ผ์ด๋ฒŒ"์˜ ์ด์•ผ๊ธฐ๋Š” ์‚ฌ๋ž‘, ์ถฉ์„ฑ์‹ฌ, ์ „์Ÿ ์ค‘ ๊ฐœ์ธ์ด ์ง๋ฉดํ•˜๋Š” ์–ด๋ ค์šด ์„ ํƒ์ด๋ผ๋Š” ์ฃผ์ œ๋ฅผ ํƒ๊ตฌํ•ฉ๋‹ˆ๋‹ค.
21. "์‹ ์„ฑํ•œ ๋ผ์ด๋ฒŒ"์˜ ์ด์•ผ๊ธฐ๋Š” ์‚ฌ๋ž‘, ์ถฉ์„ฑ์‹ฌ, ์ „์Ÿ ์ค‘ ๊ฐœ์ธ์ด ์ง๋ฉดํ•˜๋Š” ์–ด๋ ค์šด ์„ ํƒ์ด๋ผ๋Š” ์ฃผ์ œ๋ฅผ ํƒ๊ตฌํ•ฉ๋‹ˆ๋‹ค.
In this section, we provide appropriate background and methodology as it relates to continual pre-training in the context of LLMs.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ๋งฅ๋ฝ์—์„œ ์ง€์†์ ์ธ ์‚ฌ์ „ ํ•™์Šต๊ณผ ๊ด€๋ จํ•˜์—ฌ ์ ์ ˆํ•œ ๋ฐฐ๊ฒฝ ์ง€์‹๊ณผ ๋ฐฉ๋ฒ•๋ก ์„ ์ œ๊ณตํ•œ๋‹ค.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ๋งฅ๋ฝ์—์„œ ์ง€์†์ ์ธ ์‚ฌ์ „ ํ•™์Šต๊ณผ ๊ด€๋ จํ•˜์—ฌ ์ ์ ˆํ•œ ๋ฐฐ๊ฒฝ ์ง€์‹๊ณผ ๋ฐฉ๋ฒ•๋ก ์„ ์ œ๊ณตํ•œ๋‹ค.
- He might not want to share the stage with candidates he perceives as less significant, which could dilute his brand or message.
- ๊ทธ๋Š” ์ž์‹ ๋ณด๋‹ค ๋œ ์ค‘์š”ํ•˜๋‹ค๊ณ  ์ธ์‹ํ•˜๋Š” ํ›„๋ณด์ž๋“ค๊ณผ ๋ฌด๋Œ€๋ฅผ ๊ณต์œ ํ•˜๊ณ  ์‹ถ์ง€ ์•Š์„ ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋Š” ๊ทธ์˜ ๋ธŒ๋žœ๋“œ๋‚˜ ๋ฉ”์‹œ์ง€๋ฅผ ํฌ์„์‹œํ‚ฌ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ๊ทธ๋Š” ์ž์‹ ๋ณด๋‹ค ๋œ ์ค‘์š”ํ•˜๋‹ค๊ณ  ์ธ์‹ํ•˜๋Š” ํ›„๋ณด์ž๋“ค๊ณผ ๋ฌด๋Œ€๋ฅผ ๊ณต์œ ํ•˜๊ณ  ์‹ถ์ง€ ์•Š์„ ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋Š” ๊ทธ์˜ ๋ธŒ๋žœ๋“œ๋‚˜ ๋ฉ”์‹œ์ง€๋ฅผ ํฌ์„์‹œํ‚ฌ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- There could be concerns about the debate format or the choice of moderators, which might lead to a decision not to participate.
- ํ† ๋ก  ํ˜•์‹์ด๋‚˜ ์ค‘์žฌ์ž ์„ ์ •์— ๋Œ€ํ•œ ์šฐ๋ ค๊ฐ€ ์žˆ์–ด ์ฐธ์—ฌํ•˜์ง€ ์•Š๊ธฐ๋กœ ๊ฒฐ์ •ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- ํ† ๋ก  ํ˜•์‹์ด๋‚˜ ์ค‘์žฌ์ž ์„ ์ •์— ๋Œ€ํ•œ ์šฐ๋ ค๊ฐ€ ์žˆ์–ด ์ฐธ์—ฌํ•˜์ง€ ์•Š๊ธฐ๋กœ ๊ฒฐ์ •ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Figure 2: Above is the data processing flow and deduplication ratios, below is a schematic diagram of similar line deduplication.
๊ทธ๋ฆผ 2: ์œ„๋Š” ๋ฐ์ดํ„ฐ ์ฒ˜๋ฆฌ ํ๋ฆ„๊ณผ ์ค‘๋ณต ์ œ๊ฑฐ ๋น„์œจ, ์•„๋ž˜๋Š” ์œ ์‚ฌํ•œ ๋ผ์ธ ์ค‘๋ณต ์ œ๊ฑฐ์˜ ๋ชจ์‹๋„์ด๋‹ค.
๊ทธ๋ฆผ 2: ์œ„๋Š” ๋ฐ์ดํ„ฐ ์ฒ˜๋ฆฌ ํ๋ฆ„๊ณผ ์ค‘๋ณต ์ œ๊ฑฐ ๋น„์œจ, ์•„๋ž˜๋Š” ์œ ์‚ฌํ•œ ๋ผ์ธ ์ค‘๋ณต ์ œ๊ฑฐ์˜ ๋ชจ์‹๋„์ด๋‹ค.
- Generate brief search queries (up to {max_words} words each) to gather information on Google that could influence the forecast.
- ์˜ˆ์ธก์— ์˜ํ–ฅ์„ ์ค„ ์ˆ˜ ์žˆ๋Š” Google์—์„œ ์ •๋ณด๋ฅผ ์ˆ˜์ง‘ํ•˜๊ธฐ ์œ„ํ•ด ๊ฐ„๋‹จํ•œ ๊ฒ€์ƒ‰์–ด(๊ฐ๊ฐ ์ตœ๋Œ€ {max_words} ๋‹จ์–ด)๋ฅผ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค.
- ์˜ˆ์ธก์— ์˜ํ–ฅ์„ ์ค„ ์ˆ˜ ์žˆ๋Š” Google์—์„œ ์ •๋ณด๋ฅผ ์ˆ˜์ง‘ํ•˜๊ธฐ ์œ„ํ•ด ๊ฐ„๋‹จํ•œ ๊ฒ€์ƒ‰์–ด(๊ฐ๊ฐ ์ตœ๋Œ€ {์ตœ๋Œ€_๋‹จ์–ด} ๋‹จ์–ด)๋ฅผ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค.
A third scenario arises due to the automatic question generation process, some questions strongly rely on pre-existing knowledge.
์ž๋™ question ์ƒ์„ฑ ๊ณผ์ •์œผ๋กœ ์ธํ•ด ์„ธ ๋ฒˆ์งธ ์‹œ๋‚˜๋ฆฌ์˜ค๋Š” ์ผ๋ถ€ ์งˆ๋ฌธ์ด ๊ธฐ์กด ์ง€์‹์— ํฌ๊ฒŒ ์˜์กดํ•œ๋‹ค.
์ž๋™ ์งˆ๋ฌธ ์ƒ์„ฑ ๊ณผ์ •์œผ๋กœ ์ธํ•ด ์„ธ ๋ฒˆ์งธ ์‹œ๋‚˜๋ฆฌ์˜ค๋Š” ์ผ๋ถ€ ์งˆ๋ฌธ์ด ๊ธฐ์กด ์ง€์‹์— ํฌ๊ฒŒ ์˜์กดํ•œ๋‹ค.
Footnote 5: Note that according to our third assumption, the quality of the training data is uniform and independent of the size.
๊ฐ์ฃผ 5: ์šฐ๋ฆฌ์˜ ์„ธ ๋ฒˆ์งธ ๊ฐ€์ •์— ๋”ฐ๋ฅด๋ฉด, ํ›ˆ๋ จ ๋ฐ์ดํ„ฐ์˜ ํ’ˆ์งˆ์€ ๊ท ์ผํ•˜๋ฉฐ ํฌ๊ธฐ์™€ ๋ฌด๊ด€ํ•˜๋‹ค๋Š” ์ ์— ์œ ์˜ํ•˜์‹ญ์‹œ์˜ค.
๊ฐ์ฃผ 5: ์šฐ๋ฆฌ์˜ ์„ธ ๋ฒˆ์งธ ๊ฐ€์ •์— ๋”ฐ๋ฅด๋ฉด, ํ›ˆ๋ จ ๋ฐ์ดํ„ฐ์˜ ํ’ˆ์งˆ์€ ๊ท ์ผํ•˜๋ฉฐ ํฌ๊ธฐ์™€ ๋ฌด๊ด€ํ•˜๋‹ค๋Š” ์ ์— ์œ ์˜ํ•˜์‹ญ์‹œ์˜ค.
Mr. Price, from heart for their unconventional methods and reputation per unit, Jurning and Arrays include Mrs. Price, with 1984.
๊ฐ€๊ฒฉ ์”จ๋Š” ๋น„์ •ํ†ต์ ์ธ ๋ฐฉ๋ฒ•๊ณผ ๋‹จ์œ„๋‹น ๋ช…์„ฑ์œผ๋กœ ์œ ๋ช…ํ•œ ์ €๋‹๊ณผ ๋ฐฐ์—ด์„ ํฌํ•จํ•˜์—ฌ 1984๋…„ ๊ฐ€๊ฒฉ ๋ถ€์ธ์„ ํฌํ•จํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
๊ฐ€๊ฒฉ ์”จ๋Š” ์‹ฌ์žฅ์—์„œ ๋น„์ •ํ†ต์ ์ธ ๋ฐฉ๋ฒ•๊ณผ ๋‹จ์œ„๋‹น ๋ช…์„ฑ์œผ๋กœ ์œ ๋ช…ํ•œ ์ €๋‹๊ณผ ๋ฐฐ์—ด์„ ํฌํ•จํ•˜์—ฌ 1984๋…„ ๊ฐ€๊ฒฉ ๋ถ€์ธ์„ ํฌํ•จํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
Additionally, we computed the impact of a tokenizer on the average computational costs of a given model per word during training.
๋˜ํ•œ, ์šฐ๋ฆฌ๋Š” tokenizer๊ฐ€ ํ›ˆ๋ จ ์ค‘ ์ฃผ์–ด์ง„ ๋ชจ๋ธ์˜ ๋‹จ์–ด๋‹น ํ‰๊ท  ๊ณ„์‚ฐ ๋น„์šฉ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ๊ณ„์‚ฐํ–ˆ๋‹ค.
๋˜ํ•œ, ์šฐ๋ฆฌ๋Š” ํ† ํฐํ™”๊ฐ€ ํ›ˆ๋ จ ์ค‘ ์ฃผ์–ด์ง„ ๋ชจ๋ธ์˜ ๋‹จ์–ด๋‹น ํ‰๊ท  ๊ณ„์‚ฐ ๋น„์šฉ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ๊ณ„์‚ฐํ–ˆ๋‹ค.
8. Iris Winnow's bond with Roman Kitt deepens when Roman anonymously receives and responds to Iris's letters intended for Forest.
8. ๋กœ๋งŒ ํ‚คํŠธ๊ฐ€ ์•„์ด๋ฆฌ์Šค ์œˆ๋…ธ์šฐ์™€์˜ ์œ ๋Œ€๊ฐ์ด ๊นŠ์–ด์ง€๋ฉด์„œ ๋กœ๋งŒ์€ ํฌ๋ ˆ์ŠคํŠธ๋ฅผ ์œ„ํ•œ ์•„์ด๋ฆฌ์Šค์˜ ํŽธ์ง€๋ฅผ ์ต๋ช…์œผ๋กœ ๋ฐ›๊ณ  ์‘๋‹ตํ•ฉ๋‹ˆ๋‹ค.
8. ๋กœ๋งŒ ํ‚คํŠธ๊ฐ€ ์•„์ด๋ฆฌ์Šค ์œˆ๋…ธ์šฐ์™€์˜ ์œ ๋Œ€๊ฐ์ด ๊นŠ์–ด์ง€๋ฉด์„œ ๋กœ๋งŒ์€ ์ˆฒ์„ ์œ„ํ•œ ์•„์ด๋ฆฌ์Šค์˜ ํŽธ์ง€๋ฅผ ์ต๋ช…์œผ๋กœ ๋ฐ›๊ณ  ์‘๋‹ตํ•ฉ๋‹ˆ๋‹ค.
Footnote 1: Note that a LLM is not necessarily more capable than a small PLM, and emergent abilities may not occur in some LLMs.
๊ฐ์ฃผ 1: LLM์ด ๋ฐ˜๋“œ์‹œ ์†Œํ˜• PLM๋ณด๋‹ค ๋” ๋›ฐ์–ด๋‚œ ๋Šฅ๋ ฅ์„ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๊ฒƒ์€ ์•„๋‹ˆ๋ฉฐ, ์ผ๋ถ€ LLM์—์„œ๋Š” ์ƒˆ๋กœ์šด ๋Šฅ๋ ฅ์ด ๋‚˜ํƒ€๋‚˜์ง€ ์•Š์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
๊ฐ์ฃผ 1: LLM์ด ๋ฐ˜๋“œ์‹œ ์†Œํ˜• PLM๋ณด๋‹ค ๋” ๋›ฐ์–ด๋‚œ ๋Šฅ๋ ฅ์„ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๊ฒƒ์€ ์•„๋‹ˆ๋ฉฐ, ์ผ๋ถ€ LLM์—์„œ๋Š” ์ƒˆ๋กœ์šด ๋Šฅ๋ ฅ์ด ๋‚˜ํƒ€๋‚˜์ง€ ์•Š์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Figure 8: Gradient statistics over the training of a 0.2B model using WSD LRS. The exponential decay stage begins at 8000 steps.
๊ทธ๋ฆผ 8: WSD LRS๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ 0.2B ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋Š” ๋™์•ˆ์˜ Gradient ํ†ต๊ณ„. ์ง€์ˆ˜ ๊ฐ์†Œ ๋‹จ๊ณ„๋Š” 8000๋‹จ๊ณ„์—์„œ ์‹œ์ž‘๋ฉ๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 8: WSD LRS๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ 0.2B ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋Š” ๋™์•ˆ์˜ ๊ธฐ์šธ๊ธฐ ํ†ต๊ณ„. ์ง€์ˆ˜ ๊ฐ์†Œ ๋‹จ๊ณ„๋Š” 8000๋‹จ๊ณ„์—์„œ ์‹œ์ž‘๋ฉ๋‹ˆ๋‹ค.
**Question:** If \(f(x)=\frac{3x-2}{x-2}\), what is the value of \(f(-2)+f(-1)+f(0)\)? Express your answer as a common fraction.
**์งˆ๋ฌธ:** \(f(x)=\frac{3x-2}{x-2}\)์ผ ๋•Œ, \(f(-2)+f(-1)+f(0)\)์˜ ๊ฐ’์€ ์–ผ๋งˆ์ž…๋‹ˆ๊นŒ? ๋‹ต์„ ๊ณตํ†ต ๋ถ„์ˆ˜๋กœ ํ‘œํ˜„ํ•˜์„ธ์š”.
**์งˆ๋ฌธ:** f(x)=\frac{3x-2}{x-2}\)์ผ ๋•Œ, \(f(-2)+f(-1)+f(0)\)์˜ ๊ฐ’์€ ์–ผ๋งˆ์ž…๋‹ˆ๊นŒ? ๋‹ต์„ ๊ณตํ†ต ๋ถ„์ˆ˜๋กœ ํ‘œํ˜„ํ•˜์„ธ์š”.
Footnote 2: Unfortunately, we deleted the logprobs for GPT-3.5 models before we created this attack due to security constraints.
๊ฐ์ฃผ 2: ์•ˆํƒ€๊น๊ฒŒ๋„ ๋ณด์•ˆ ์ œ์•ฝ์œผ๋กœ ์ธํ•ด ์ด ๊ณต๊ฒฉ์„ ๋งŒ๋“ค๊ธฐ ์ „์— GPT-3.5 ๋ชจ๋ธ์— ๋Œ€ํ•œ logprobs๋ฅผ ์‚ญ์ œํ–ˆ์Šต๋‹ˆ๋‹ค.
๊ฐ์ฃผ 2: ์•ˆํƒ€๊น๊ฒŒ๋„ ๋ณด์•ˆ ์ œ์•ฝ์œผ๋กœ ์ธํ•ด ์ด ๊ณต๊ฒฉ์„ ๋งŒ๋“ค๊ธฐ ์ „์— GPT-3.5 ๋ชจ๋ธ์— ๋Œ€ํ•œ ๋กœ๊ทธ ํ™•๋ฅ ์„ ์‚ญ์ œํ–ˆ์Šต๋‹ˆ๋‹ค.
Furthermore, a recent study [276] defines the distance required to rotate one cycle (\(2\pi\)) for each dimension as wavelength:
๋˜ํ•œ ์ตœ๊ทผ ์—ฐ๊ตฌ [276]์—์„œ๋Š” ๊ฐ ์ฐจ์›์— ๋Œ€ํ•ด ํ•˜๋‚˜์˜ ์ฃผ๊ธฐ(\(2\pi\))๋ฅผ ํšŒ์ „ํ•˜๋Š” ๋ฐ ํ•„์š”ํ•œ ๊ฑฐ๋ฆฌ๋ฅผ wavelength๋กœ ์ •์˜ํ•œ๋‹ค.
๋˜ํ•œ ์ตœ๊ทผ ์—ฐ๊ตฌ [276]์—์„œ๋Š” ๊ฐ ์ฐจ์›์— ๋Œ€ํ•ด ํ•˜๋‚˜์˜ ์ฃผ๊ธฐ(\(2\pi\))๋ฅผ ํšŒ์ „ํ•˜๋Š” ๋ฐ ํ•„์š”ํ•œ ๊ฑฐ๋ฆฌ๋ฅผ ํŒŒ์žฅ์œผ๋กœ ์ •์˜ํ•œ๋‹ค.
PersonA: The main ingredients are typically chile peppers, garlic, and vinegar. However, there can be variations in the recipe.
์‚ฌ๋žŒA: ์ฃผ์š” ์žฌ๋ฃŒ๋Š” ์ผ๋ฐ˜์ ์œผ๋กœ ๊ณ ์ถ”, ๋งˆ๋Š˜, ์‹์ดˆ์ž…๋‹ˆ๋‹ค. ํ•˜์ง€๋งŒ ๋ ˆ์‹œํ”ผ์— ๋ณ€ํ˜•์ด ์žˆ์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
์‚ฌ๋žŒA: ์ฃผ์š” ์žฌ๋ฃŒ๋Š” ์ผ๋ฐ˜์ ์œผ๋กœ ๊ณ ์ถ”, ๋งˆ๋Š˜, ์‹์ดˆ์ž…๋‹ˆ๋‹ค. ํ•˜์ง€๋งŒ ๋ ˆ์‹œํ”ผ์— ๋ณ€ํ˜•์ด ์žˆ์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
and then convert the loss minimization problem from Equation 2 into the following negative log-likelihood minimization problem:
๊ทธ๋ฆฌ๊ณ  ๋ฐฉ์ •์‹ 2์˜ Loss ์ตœ์†Œํ™” ๋ฌธ์ œ๋ฅผ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์Œ์˜ ๋กœ๊ทธ ๊ฐ€๋Šฅ์„ฑ ์ตœ์†Œํ™” ๋ฌธ์ œ๋กœ ๋ณ€ํ™˜ํ•ฉ๋‹ˆ๋‹ค:
๊ทธ๋ฆฌ๊ณ  ๋ฐฉ์ •์‹ 2์˜ ์†์‹ค ์ตœ์†Œํ™” ๋ฌธ์ œ๋ฅผ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์Œ์˜ ๋กœ๊ทธ ๊ฐ€๋Šฅ์„ฑ ์ตœ์†Œํ™” ๋ฌธ์ œ๋กœ ๋ณ€ํ™˜ํ•ฉ๋‹ˆ๋‹ค:
AI Agent Agent Architecture AI Reasoning Planning Tool Calling Single Agent Multi Agent Agent Survey LLM Agent Autonomous Agent
AI Agent ์—์ด์ „ํŠธ ์•„ํ‚คํ…์ฒ˜ AI ์ถ”๋ก  ๊ณ„ํš ๋„๊ตฌ ํ˜ธ์ถœ ๋‹จ์ผ ์—์ด์ „ํŠธ ๋‹ค์ค‘ ์—์ด์ „ํŠธ ์—์ด์ „ํŠธ ์„ค๋ฌธ์กฐ์‚ฌ LLM Agent ์ž์œจ ์—์ด์ „ํŠธ
AI ์—์ด์ „ํŠธ ์—์ด์ „ํŠธ ์•„ํ‚คํ…์ฒ˜ AI ์ถ”๋ก  ๊ณ„ํš ๋„๊ตฌ ํ˜ธ์ถœ ๋‹จ์ผ ์—์ด์ „ํŠธ ๋‹ค์ค‘ ์—์ด์ „ํŠธ ์—์ด์ „ํŠธ ์„ค๋ฌธ์กฐ์‚ฌ LLM ์—์ด์ „ํŠธ ์ž์œจ ์—์ด์ „ํŠธ
DPO is relatively straightforward to implement; PyTorch code for the DPO loss is provided below:import torch.nn.functional as F
DPO๋Š” ๊ตฌํ˜„์ด ๋น„๊ต์  ๊ฐ„๋‹จํ•˜๋ฉฐ, ์•„๋ž˜๋Š” DPO Loss์— ๋Œ€ํ•œ PyTorch ์ฝ”๋“œ์ž…๋‹ˆ๋‹ค:import torch.nn.functional as F
DPO๋Š” ๊ตฌํ˜„์ด ๋น„๊ต์  ๊ฐ„๋‹จํ•˜๋ฉฐ, ์•„๋ž˜๋Š” DPO ์†์‹ค์— ๋Œ€ํ•œ PyTorch ์ฝ”๋“œ์ž…๋‹ˆ๋‹ค:import torch.nn.functional as F
Footnote 3: The impact of using different evaluation models, such as only using XCOMET or KIWI-XXL, is explored in Section 5.1.
๊ฐ์ฃผ 3: XCOMET ๋˜๋Š” KIWI-XXL๋งŒ ์‚ฌ์šฉํ•˜๋Š” ๋“ฑ ๋‹ค์–‘ํ•œ evaluation ๋ชจ๋ธ์„ ์‚ฌ์šฉํ–ˆ์„ ๋•Œ์˜ ์˜ํ–ฅ์€ ์„น์…˜ 5.1์—์„œ ์‚ดํŽด๋ณธ๋‹ค.
๊ฐ์ฃผ 3: XCOMET ๋˜๋Š” KIWI-XXL๋งŒ ์‚ฌ์šฉํ•˜๋Š” ๋“ฑ ๋‹ค์–‘ํ•œ ํ‰๊ฐ€ ๋ชจ๋ธ์„ ์‚ฌ์šฉํ–ˆ์„ ๋•Œ์˜ ์˜ํ–ฅ์€ ์„น์…˜ 5.1์—์„œ ์‚ดํŽด๋ณธ๋‹ค.
(b) The second search query prompt we use. It first asks the model to consider sub-questions and use that to steer the outputs.
(b) ๋‘ ๋ฒˆ์งธ ๊ฒ€์ƒ‰ ์ฟผ๋ฆฌ prompt์ž…๋‹ˆ๋‹ค. ๋จผ์ € ๋ชจ๋ธ์— ํ•˜์œ„ ์งˆ๋ฌธ์„ ๊ณ ๋ คํ•˜๊ณ  ์ด๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ถœ๋ ฅ์„ ์œ ๋„ํ•˜๋„๋ก ์š”์ฒญํ•ฉ๋‹ˆ๋‹ค.
(b) ๋‘ ๋ฒˆ์งธ ๊ฒ€์ƒ‰ ์ฟผ๋ฆฌ ํ”„๋กฌํ”„ํŠธ์ž…๋‹ˆ๋‹ค. ๋จผ์ € ๋ชจ๋ธ์— ํ•˜์œ„ ์งˆ๋ฌธ์„ ๊ณ ๋ คํ•˜๊ณ  ์ด๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ถœ๋ ฅ์„ ์œ ๋„ํ•˜๋„๋ก ์š”์ฒญํ•ฉ๋‹ˆ๋‹ค.
Using Eqn. 8, we can fit the loss on each validation domain \(L_{i}\) and sum them up to obtain the prediction of overall loss.
์‹ 8์„ ์‚ฌ์šฉํ•˜์—ฌ ๊ฐ ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ๋„๋ฉ”์ธ \(L_{i}\)์˜ Loss๋ฅผ ๋งž์ถ”๊ณ  ํ•ฉ์‚ฐํ•˜์—ฌ ์ „์ฒด Loss์˜ ์˜ˆ์ธก์„ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค.
์‹ 8์„ ์‚ฌ์šฉํ•˜์—ฌ ๊ฐ ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ๋„๋ฉ”์ธ \(L_{i}\)์˜ ์†์‹ค์„ ๋งž์ถ”๊ณ  ํ•ฉ์‚ฐํ•˜์—ฌ ์ „์ฒด ์†์‹ค์˜ ์˜ˆ์ธก์„ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค.