question
stringlengths
101
2.69k
chosen
stringlengths
42
2.7k
rejected
stringlengths
42
2.18k
To explicitly explore the impact of these dimensions, we conduct a mathematical task designed to isolate each one.
์ด๋Ÿฌํ•œ ์ฐจ์›์˜ ์˜ํ–ฅ์„ ๋ช…์‹œ์ ์œผ๋กœ ํƒ๊ตฌํ•˜๊ธฐ ์œ„ํ•ด ๊ฐ ์ฐจ์›์„ ๋ถ„๋ฆฌํ•˜๋„๋ก ์„ค๊ณ„๋œ ์ˆ˜ํ•™์  ๊ณผ์ œ๋ฅผ ์ˆ˜ํ–‰ํ•œ๋‹ค.
์ด๋Ÿฌํ•œ ์ฐจ์›์˜ ์˜ํ–ฅ์„ ๋ช…์‹œ์ ์œผ๋กœ ํƒ๊ตฌํ•˜๊ธฐ ์œ„ํ•ด ๊ฐ ์ฐจ์›์„ ๋ถ„๋ฆฌํ•˜๋„๋ก ์„ค๊ณ„๋œ ์ˆ˜ํ•™์  ๊ณผ์ œ๋ฅผ ์ˆ˜ํ–‰ํ•œ๋‹ค.
As the raw system is continuous, we need to first discretize them before feeding the computer, as shown in Fig. 2.
์›์‹œ ์‹œ์Šคํ…œ์ด ์—ฐ์†์ ์ด๊ธฐ ๋•Œ๋ฌธ์— ์ปดํ“จํ„ฐ์— ๊ณต๊ธ‰ํ•˜๊ธฐ ์ „์— ๋จผ์ € ์ด๋ฅผ discretizeํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค(๊ทธ๋ฆผ 2 ์ฐธ์กฐ).
์›์‹œ ์‹œ์Šคํ…œ์ด ์—ฐ์†์ ์ด๊ธฐ ๋•Œ๋ฌธ์— ์ปดํ“จํ„ฐ์— ๊ณต๊ธ‰ํ•˜๊ธฐ ์ „์— ๋จผ์ € ์ด๋ฅผ ๋ถˆ์—ฐ์†ํ™”ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค(๊ทธ๋ฆผ 2 ์ฐธ์กฐ).
Dataset SizesTable 13 gives the size in BPE tokens and in examples of each dataset before and after deduplication.
๋ฐ์ดํ„ฐ ์„ธํŠธ ํฌ๊ธฐํ‘œ 13์€ ์ค‘๋ณต ์ œ๊ฑฐ ์ „ํ›„์˜ ๊ฐ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ BPE ํ† ํฐ ๋ฐ ์˜ˆ์ œ ์ˆ˜์˜ ํฌ๊ธฐ๋ฅผ ๋‚˜ํƒ€๋‚ธ๋‹ค.
๋ฐ์ดํ„ฐ ์„ธํŠธ ํฌ๊ธฐํ‘œ 13์€ ์ค‘๋ณต ์ œ๊ฑฐ ์ „ํ›„์˜ ๊ฐ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ BPE ํ† ํฐ ๋ฐ ์˜ˆ์ œ ์ˆ˜์˜ ํฌ๊ธฐ๋ฅผ ๋‚˜ํƒ€๋‚ธ๋‹ค.
Thus, summing up over all the diversity part, we have (recall we are fixing \(Q\) and thus fixing \(\mathcal{Z}\))
๋”ฐ๋ผ์„œ ๋ชจ๋“  ๋‹ค์–‘์„ฑ ๋ถ€๋ถ„์„ ํ•ฉํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค(Q๋ฅผ ์ˆ˜์ •ํ•˜๊ณ  ์žˆ์œผ๋ฏ€๋กœ \(\mathcal{Z}\)๋ฅผ ์ˆ˜์ •ํ•˜๊ณ  ์žˆ๋‹ค๋Š” ์ ์„ ๊ธฐ์–ตํ•˜์„ธ์š”).
๋”ฐ๋ผ์„œ ๋ชจ๋“  ๋‹ค์–‘์„ฑ ๋ถ€๋ถ„์„ ํ•ฉํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค(Q๋ฅผ ์ˆ˜์ •ํ•˜๊ณ  ์žˆ์œผ๋ฏ€๋กœ \(\mathcal{Z}\)๋ฅผ ์ˆ˜์ •ํ•˜๊ณ  ์žˆ๋‹ค๋Š” ์ ์„ ๊ธฐ์–ตํ•˜์„ธ์š”).
Typically, domain-specific LLMs are built by training the model from scratch using massive amounts of domain data.
์ผ๋ฐ˜์ ์œผ๋กœ ๋„๋ฉ”์ธ๋ณ„ LLM์€ ๋ฐฉ๋Œ€ํ•œ ์–‘์˜ ๋„๋ฉ”์ธ ๋ฐ์ดํ„ฐ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ์ฒ˜์Œ๋ถ€ํ„ฐ ํ•™์Šต์‹œ์ผœ ๊ตฌ์ถ•ํ•ฉ๋‹ˆ๋‹ค.
์ผ๋ฐ˜์ ์œผ๋กœ ๋„๋ฉ”์ธ๋ณ„ LLM์€ ๋ฐฉ๋Œ€ํ•œ ์–‘์˜ ๋„๋ฉ”์ธ ๋ฐ์ดํ„ฐ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ์ฒ˜์Œ๋ถ€ํ„ฐ ํ•™์Šต์‹œ์ผœ ๊ตฌ์ถ•ํ•ฉ๋‹ˆ๋‹ค.
The fitted curve along the data axis for each dataset and each checkpoints are shown in orange lines in Figure 17.
๊ฐ ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ๊ณผ ๊ฐ ์ฒดํฌํฌ์ธํŠธ์— ๋Œ€ํ•œ ๋ฐ์ดํ„ฐ ์ถ•์„ ๋”ฐ๋ผ ๋งž์ถค ๊ณก์„ ์€ ๊ทธ๋ฆผ 17์˜ ์ฃผํ™ฉ์ƒ‰ ์„ ์œผ๋กœ ํ‘œ์‹œ๋œ๋‹ค.
๊ฐ ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ๊ณผ ๊ฐ ์ฒดํฌํฌ์ธํŠธ์— ๋Œ€ํ•œ ๋ฐ์ดํ„ฐ ์ถ•์„ ๋”ฐ๋ผ ๋งž์ถค ๊ณก์„ ์€ ๊ทธ๋ฆผ 17์˜ ์ฃผํ™ฉ์ƒ‰ ์„ ์œผ๋กœ ํ‘œ์‹œ๋œ๋‹ค.
FindingsThe results in Tab. 1 suggests both M1 and M4 gives reliable predictions while M4 has fewer coefficients.
๊ฒฐ๊ณผ Tab. 1์˜ ๊ฒฐ๊ณผ๋Š” M1๊ณผ M4๊ฐ€ ๋ชจ๋‘ ์‹ ๋ขฐํ•  ์ˆ˜ ์žˆ๋Š” ์˜ˆ์ธก์„ ์ œ๊ณตํ•˜์ง€๋งŒ M4์˜ ๊ณ„์ˆ˜๊ฐ€ ๋” ์ ๋‹ค๋Š” ๊ฒƒ์„ ์‹œ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
๊ฒฐ๊ณผTab. 1์˜ ๊ฒฐ๊ณผ๋Š” M1๊ณผ M4๊ฐ€ ๋ชจ๋‘ ์‹ ๋ขฐํ•  ์ˆ˜ ์žˆ๋Š” ์˜ˆ์ธก์„ ์ œ๊ณตํ•˜์ง€๋งŒ M4์˜ ๊ณ„์ˆ˜๊ฐ€ ๋” ์ ๋‹ค๋Š” ๊ฒƒ์„ ์‹œ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
1. Unstable Anjnana 2. Acute Myocardial Infarction 3. Stable Anjnana 4. Gastroesophageal Reflux Disease (GERD) 5.
1. ๋ถˆ์•ˆ์ • ์•ˆ์ฆˆ๋‚˜ 2. ๊ธ‰์„ฑ ์‹ฌ๊ทผ๊ฒฝ์ƒ‰์ฆ 3. ์•ˆ์ • ์•ˆ์ฆˆ๋‚˜ 4. ์œ„์‹๋„ ์—ญ๋ฅ˜ ์งˆํ™˜(GERD) 5.
1. ๋ถˆ์•ˆ์ • ์•ˆ์ฆˆ๋‚˜2. ๊ธ‰์„ฑ ์‹ฌ๊ทผ๊ฒฝ์ƒ‰์ฆ3. ์•ˆ์ • ์•ˆ์ฆˆ๋‚˜4. ์œ„์‹๋„ ์—ญ๋ฅ˜ ์งˆํ™˜(GERD)5.
Derivation. Based on Property 3, we propose hierarchical sampling-and-voting can further enhance the performance.
ํŒŒ์ƒ. ์†์„ฑ 3์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๊ณ„์ธต์  ์ƒ˜ํ”Œ๋ง ๋ฐ ํˆฌํ‘œ๋ฅผ ํ†ตํ•ด ์„ฑ๋Šฅ์„ ๋”์šฑ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‹ค๊ณ  ์ œ์•ˆํ•ฉ๋‹ˆ๋‹ค.
ํŒŒ์ƒ. ์†์„ฑ 3์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๊ณ„์ธต์  ์ƒ˜ํ”Œ๋ง ๋ฐ ํˆฌํ‘œ๋ฅผ ํ†ตํ•ด ์„ฑ๋Šฅ์„ ๋”์šฑ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‹ค๊ณ  ์ œ์•ˆํ•ฉ๋‹ˆ๋‹ค.
In the case of tail narrowing, the scaling behavior changes; instead of a plateau, we obtain a slower decay rate:
๊ผฌ๋ฆฌ ์ขํžˆ๊ธฐ์˜ ๊ฒฝ์šฐ, scaling ๋™์ž‘์ด ๋ณ€ํ™”ํ•˜์—ฌ, ๊ณ ์› ๋Œ€์‹ ์— ๋Š๋ฆฐ ๊ฐ์‡ ์œจ์„ ์–ป๊ฒŒ ๋œ๋‹ค.
๊ผฌ๋ฆฌ ์ขํžˆ๊ธฐ์˜ ๊ฒฝ์šฐ, ์Šค์ผ€์ผ๋ง ๋™์ž‘์ด ๋ณ€ํ™”ํ•˜์—ฌ, ๊ณ ์› ๋Œ€์‹ ์— ๋Š๋ฆฐ ๊ฐ์‡ ์œจ์„ ์–ป๊ฒŒ ๋œ๋‹ค.
Now, to find the total number of gems in the chest, we add the number of diamonds, rubies, and emeralds together.
์ด์ œ ์ƒ์ž์— ์žˆ๋Š” ์ด ๋ณด์„ ์ˆ˜๋ฅผ ๊ตฌํ•˜๊ธฐ ์œ„ํ•ด ๋‹ค์ด์•„๋ชฌ๋“œ, ๋ฃจ๋น„, ์—๋ฉ”๋ž„๋“œ์˜ ์ˆ˜๋ฅผ ๋”ํ•ฉ๋‹ˆ๋‹ค.
์ด์ œ ์ƒ์ž์— ์žˆ๋Š” ์ด ๋ณด์„ ์ˆ˜๋ฅผ ๊ตฌํ•˜๊ธฐ ์œ„ํ•ด ๋‹ค์ด์•„๋ชฌ๋“œ, ๋ฃจ๋น„, ์—๋ฉ”๋ž„๋“œ์˜ ์ˆ˜๋ฅผ ๋”ํ•ฉ๋‹ˆ๋‹ค.
There are many frameworks developed for optimized training of LLMs, here we introduce some of the prominent ones.
LLM์˜ ์ตœ์ ํ™”๋œ ํ›ˆ๋ จ์„ ์œ„ํ•ด ๊ฐœ๋ฐœ๋œ ๋งŽ์€ ํ”„๋ ˆ์ž„์›Œํฌ๊ฐ€ ์žˆ์œผ๋ฉฐ, ์—ฌ๊ธฐ์—์„œ๋Š” ์ฃผ๋ชฉํ•  ๋งŒํ•œ ๋ช‡ ๊ฐ€์ง€๋ฅผ ์†Œ๊ฐœํ•œ๋‹ค.
LLM์˜ ์ตœ์ ํ™”๋œ ํ›ˆ๋ จ์„ ์œ„ํ•ด ๊ฐœ๋ฐœ๋œ ๋งŽ์€ ํ”„๋ ˆ์ž„์›Œํฌ๊ฐ€ ์žˆ์œผ๋ฉฐ, ์—ฌ๊ธฐ์—์„œ๋Š” ์ฃผ๋ชฉํ•  ๋งŒํ•œ ๋ช‡ ๊ฐ€์ง€๋ฅผ ์†Œ๊ฐœํ•œ๋‹ค.
To find the value of \(f(-2)+f(-1)+f(0)\), we first need to evaluate each function at these specific values of x.
f(-2)+f(-1)+f(0)์˜ ๊ฐ’์„ ๊ตฌํ•˜๋ ค๋ฉด ๋จผ์ € ๊ฐ ํ•จ์ˆ˜๋ฅผ x์˜ ํŠน์ • ๊ฐ’์—์„œ ํ‰๊ฐ€ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
f(-2)+f(-1)+f(0)\)์˜ ๊ฐ’์„ ๊ตฌํ•˜๋ ค๋ฉด ๋จผ์ € ๊ฐ ํ•จ์ˆ˜๋ฅผ x์˜ ํŠน์ • ๊ฐ’์—์„œ ํ‰๊ฐ€ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
**Methods.** Due to cost constraints, we cannot afford to evaluate relevance scores using the full article texts.
**๋ฐฉ๋ฒ•.** ๋น„์šฉ ์ œ์•ฝ์œผ๋กœ ์ธํ•ด ์ „์ฒด ๊ธฐ์‚ฌ ํ…์ŠคํŠธ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๊ด€๋ จ์„ฑ ์ ์ˆ˜๋ฅผ ํ‰๊ฐ€ํ•  ์—ฌ๋ ฅ์ด ์—†์Šต๋‹ˆ๋‹ค.
**๋ฐฉ๋ฒ•.** ๋น„์šฉ ์ œ์•ฝ์œผ๋กœ ์ธํ•ด ์ „์ฒด ๊ธฐ์‚ฌ ํ…์ŠคํŠธ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๊ด€๋ จ์„ฑ ์ ์ˆ˜๋ฅผ ํ‰๊ฐ€ํ•  ์—ฌ๋ ฅ์ด ์—†์Šต๋‹ˆ๋‹ค.
Motivated by this, we briefly experiment with different embedding spaces and discuss our results in this section.
์ด์— ๋™๊ธฐ๋ฅผ ๋ถ€์—ฌํ•˜์—ฌ, ๋ณธ ์ ˆ์—์„œ๋Š” ๋‹ค์–‘ํ•œ embedding ๊ณต๊ฐ„์— ๋Œ€ํ•ด ๊ฐ„๋žตํ•˜๊ฒŒ ์‹คํ—˜์„ ์ง„ํ–‰ํ•˜๊ณ  ๊ทธ ๊ฒฐ๊ณผ์— ๋Œ€ํ•ด ๋…ผ์˜ํ•œ๋‹ค.
์ด์— ๋™๊ธฐ๋ฅผ ๋ถ€์—ฌํ•˜์—ฌ, ๋ณธ ์ ˆ์—์„œ๋Š” ๋‹ค์–‘ํ•œ ์ž„๋ฒ ๋”ฉ ๊ณต๊ฐ„์— ๋Œ€ํ•ด ๊ฐ„๋žตํ•˜๊ฒŒ ์‹คํ—˜์„ ์ง„ํ–‰ํ•˜๊ณ  ๊ทธ ๊ฒฐ๊ณผ์— ๋Œ€ํ•ด ๋…ผ์˜ํ•œ๋‹ค.
Fig. 13: Relative Response Quality of Vicuna and a few other well-known models by GPT-4. Courtesy of Vicuna Team.
๊ทธ๋ฆผ 13: GPT-4์— ์˜ํ•œ Vicuna ๋ฐ ๊ธฐํƒ€ ์ž˜ ์•Œ๋ ค์ง„ ๋ช‡ ๊ฐ€์ง€ ๋ชจ๋ธ์˜ ์ƒ๋Œ€์  ์‘๋‹ต ํ’ˆ์งˆ. Vicuna ํŒ€์˜ ํ˜‘์กฐ.
๊ทธ๋ฆผ 13: GPT-4์— ์˜ํ•œ ๋น„์ฟ ๋‚˜ ๋ฐ ๊ธฐํƒ€ ์ž˜ ์•Œ๋ ค์ง„ ๋ช‡ ๊ฐ€์ง€ ๋ชจ๋ธ์˜ ์ƒ๋Œ€์  ์‘๋‹ต ํ’ˆ์งˆ. Vicuna ํŒ€์˜ ํ˜‘์กฐ.
Figure 4: Best of \(N\) baseline for \(N=\{1,4,16,64,128\}\). Performance plateaus after roughly 64-128 samples.
๊ทธ๋ฆผ 4: \(N=\{1,4,16,64,128\}\)์— ๋Œ€ํ•œ ์ตœ๊ณ ์˜ \(N\) baseline. ์•ฝ 64-128๊ฐœ์˜ ์ƒ˜ํ”Œ ์ดํ›„ ์„ฑ๋Šฅ์ด ์ •์ฒด๋ฉ๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 4: N=\{1,4,16,64,128\}\)์— ๋Œ€ํ•œ ์ตœ๊ณ ์˜ \(N\) ๊ธฐ์ค€์„ . ์•ฝ 64-128๊ฐœ์˜ ์ƒ˜ํ”Œ ์ดํ›„ ์„ฑ๋Šฅ์ด ์ •์ฒด๋ฉ๋‹ˆ๋‹ค.
**Question:** What environmental concern was raised by some scientists as a result of the 2023 Hawaii wildfires?
**์งˆ๋ฌธ:** 2023๋…„ ํ•˜์™€์ด ์‚ฐ๋ถˆ๋กœ ์ธํ•ด ์ผ๋ถ€ ๊ณผํ•™์ž๋“ค์ด ์ œ๊ธฐํ•œ ํ™˜๊ฒฝ ๋ฌธ์ œ๋Š” ๋ฌด์—‡์ธ๊ฐ€์š”?
**์งˆ๋ฌธ:** 2023๋…„ ํ•˜์™€์ด ์‚ฐ๋ถˆ๋กœ ์ธํ•ด ์ผ๋ถ€ ๊ณผํ•™์ž๋“ค์ด ์ œ๊ธฐํ•œ ํ™˜๊ฒฝ ๋ฌธ์ œ๋Š” ๋ฌด์—‡์ธ๊ฐ€์š”?
**Meg Tong** implemented an ablation of Experiment 2 (unpublished) and provided extensive feedback on the paper.
**Meg Tong**์€ ์‹คํ—˜ 2์˜ ablation(๋ฏธ๋ฐœํ‘œ)์„ ๊ตฌํ˜„ํ•˜๊ณ  ๋…ผ๋ฌธ์— ๋Œ€ํ•œ ๊ด‘๋ฒ”์œ„ํ•œ ํ”ผ๋“œ๋ฐฑ์„ ์ œ๊ณตํ–ˆ์Šต๋‹ˆ๋‹ค.
**Meg Tong**์€ ์‹คํ—˜ 2์˜ ์ ˆ์ œ์ˆ (๋ฏธ๋ฐœํ‘œ)์„ ๊ตฌํ˜„ํ•˜๊ณ  ๋…ผ๋ฌธ์— ๋Œ€ํ•œ ๊ด‘๋ฒ”์œ„ํ•œ ํ”ผ๋“œ๋ฐฑ์„ ์ œ๊ณตํ–ˆ์Šต๋‹ˆ๋‹ค.
12. Iris maintains a magical correspondence with a mysterious figure named Carver through enchanted typewriters.
์•„์ด๋ฆฌ์Šค๋Š” ๋งˆ๋ฒ•์— ๊ฑธ๋ฆฐ ํƒ€์ž๊ธฐ๋ฅผ ํ†ตํ•ด Carver๋ผ๋Š” ์‹ ๋น„ํ•œ ์ธ๋ฌผ๊ณผ ๋งˆ๋ฒ•์˜ ํŽธ์ง€๋ฅผ ์ฃผ๊ณ ๋ฐ›์Šต๋‹ˆ๋‹ค.
12. ์•„์ด๋ฆฌ์Šค๋Š” ๋งˆ๋ฒ•์— ๊ฑธ๋ฆฐ ํƒ€์ž๊ธฐ๋ฅผ ํ†ตํ•ด ์นด๋ฒ„๋ผ๋Š” ์‹ ๋น„ํ•œ ์ธ๋ฌผ๊ณผ ๋งˆ๋ฒ•์˜ ํŽธ์ง€๋ฅผ ์ฃผ๊ณ ๋ฐ›์Šต๋‹ˆ๋‹ค.
We leveraged OpenAI API's function calling feature to ensure that the model would response with a strict format.
๋ชจ๋ธ์ด ์—„๊ฒฉํ•œ ํ˜•์‹์œผ๋กœ ์‘๋‹ตํ•˜๋„๋ก ํ•˜๊ธฐ ์œ„ํ•ด OpenAI API์˜ ํ•จ์ˆ˜ ํ˜ธ์ถœ ๊ธฐ๋Šฅ์„ ํ™œ์šฉํ–ˆ์Šต๋‹ˆ๋‹ค.
๋ชจ๋ธ์ด ์—„๊ฒฉํ•œ ํ˜•์‹์œผ๋กœ ์‘๋‹ตํ•˜๋„๋ก ํ•˜๊ธฐ ์œ„ํ•ด OpenAI API์˜ ํ•จ์ˆ˜ ํ˜ธ์ถœ ๊ธฐ๋Šฅ์„ ํ™œ์šฉํ–ˆ์Šต๋‹ˆ๋‹ค.
We conducted the same experiment in their setting, with respect to token, word, and entity-preserving reversals.
ํ† ํฐ, ๋‹จ์–ด ๋ฐ ์—”ํ‹ฐํ‹ฐ ๋ณด์กด ๋ฆฌ๋ฒ„์„ค๊ณผ ๊ด€๋ จํ•˜์—ฌ ํ•ด๋‹น ์„ค์ •์—์„œ ๋™์ผํ•œ ์‹คํ—˜์„ ์ˆ˜ํ–‰ํ–ˆ๋‹ค.
ํ† ํฐ, ๋‹จ์–ด ๋ฐ ์—”ํ‹ฐํ‹ฐ ๋ณด์กด ๋ฆฌ๋ฒ„์„ค๊ณผ ๊ด€๋ จํ•˜์—ฌ ํ•ด๋‹น ์„ค์ •์—์„œ ๋™์ผํ•œ ์‹คํ—˜์„ ์ˆ˜ํ–‰ํ–ˆ๋‹ค.
Figure 7: Prompt template to encode the input into metadata, consisting of its use case and transferable skills.
๊ทธ๋ฆผ 7: ์‚ฌ์šฉ ์‚ฌ๋ก€์™€ ์ „๋‹ฌ ๊ฐ€๋Šฅํ•œ ๊ธฐ์ˆ ๋กœ ๊ตฌ์„ฑ๋œ ์ž…๋ ฅ์„ ๋ฉ”ํƒ€๋ฐ์ดํ„ฐ๋กœ ์ธ์ฝ”๋”ฉํ•˜๋Š” prompt ํ…œํ”Œ๋ฆฟ์ž…๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 7: ์‚ฌ์šฉ ์‚ฌ๋ก€์™€ ์ „๋‹ฌ ๊ฐ€๋Šฅํ•œ ๊ธฐ์ˆ ๋กœ ๊ตฌ์„ฑ๋œ ์ž…๋ ฅ์„ ๋ฉ”ํƒ€๋ฐ์ดํ„ฐ๋กœ ์ธ์ฝ”๋”ฉํ•˜๋Š” ํ”„๋กฌํ”„ํŠธ ํ…œํ”Œ๋ฆฟ์ž…๋‹ˆ๋‹ค.
**Theorem 2.4** (Informal).: _Model collapse (as spoken of in the literature) occurs iff \(n\gg(T_{0}/T)^{c}\)._
**์ •๋ฆฌ 2.4** (๋น„๊ณต์‹).: _๋ฌธํ—Œ์—์„œ ๋งํ•˜๋Š” ๊ฒƒ์ฒ˜๋Ÿผ ๋ชจ๋ธ ๋ถ•๊ดด๋Š” \(n\gg(T_{0}/T)^{c}\)._์ผ ๊ฒฝ์šฐ์— ๋ฐœ์ƒํ•ฉ๋‹ˆ๋‹ค.
**์ •๋ฆฌ 2.4** (๋น„๊ณต์‹).: _๋ฌธํ—Œ์—์„œ ๋งํ•˜๋Š” ๊ฒƒ์ฒ˜๋Ÿผ ๋ชจ๋ธ ๋ถ•๊ดด๋Š” \(n\gg(T_{0}/T)^{c}\)._์ผ ๊ฒฝ์šฐ์— ๋ฐœ์ƒํ•ฉ๋‹ˆ๋‹ค.
This section provides an overview of the benchmarks and datasets suited to evaluate the basic abilities of LLMs.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ๊ธฐ๋ณธ ๋Šฅ๋ ฅ์„ ํ‰๊ฐ€ํ•˜๋Š” ๋ฐ ์ ํ•ฉํ•œ ๋ฒค์น˜๋งˆํฌ์™€ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ๋Œ€ํ•œ ๊ฐœ์š”๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ๊ธฐ๋ณธ ๋Šฅ๋ ฅ์„ ํ‰๊ฐ€ํ•˜๋Š” ๋ฐ ์ ํ•ฉํ•œ ๋ฒค์น˜๋งˆํฌ์™€ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ๋Œ€ํ•œ ๊ฐœ์š”๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
Figure 5: We alter the premise in the question to investigate the impact of changing the word โ€œthreeโ€ to โ€œfiveโ€.
๊ทธ๋ฆผ 5: ๋ฌธ์ œ์˜ ์ „์ œ๋ฅผ ๋ณ€๊ฒฝํ•˜์—ฌ "three"์ด๋ผ๋Š” ๋‹จ์–ด๋ฅผ "five"๋กœ ๋ณ€๊ฒฝํ–ˆ์„ ๋•Œ์˜ ์˜ํ–ฅ์„ ์กฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 5: ๋ฌธ์ œ์˜ ์ „์ œ๋ฅผ ๋ณ€๊ฒฝํ•˜์—ฌ "3"์ด๋ผ๋Š” ๋‹จ์–ด๋ฅผ "5"๋กœ ๋ณ€๊ฒฝํ–ˆ์„ ๋•Œ์˜ ์˜ํ–ฅ์„ ์กฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
For our \(\mathsf{bioS}(N)\) data, we define a slightly reduced capacity ratio by omitting the diversity term.11
์šฐ๋ฆฌ์˜ \(\mathsf{bioS}(N)\) ๋ฐ์ดํ„ฐ์˜ ๊ฒฝ์šฐ, ๋‹ค์–‘์„ฑ ํ•ญ์„ ์ƒ๋žตํ•˜์—ฌ ์•ฝ๊ฐ„ ์ถ•์†Œ๋œ ์šฉ๋Ÿ‰ ๋น„์œจ์„ ์ •์˜ํ•œ๋‹ค.11
์šฐ๋ฆฌ์˜ \(\mathsf{bioS}(N)\) ๋ฐ์ดํ„ฐ์˜ ๊ฒฝ์šฐ, ๋‹ค์–‘์„ฑ ํ•ญ์„ ์ƒ๋žตํ•˜์—ฌ ์•ฝ๊ฐ„ ์ถ•์†Œ๋œ ์šฉ๋Ÿ‰ ๋น„์œจ์„ ์ •์˜ํ•œ๋‹ค.11
Jakub Lacki, Vahab Mirrokni, and Michal Wlodarczyk. 2018. Connected components at scale via local contractions.
์•ผ์ฟ ๋ธŒ ๋ผํ‚ค, ๋ฐ”ํ•ฉ ๋ฏธ๋กœํฌ๋‹ˆ, ๋ฏธํ•  ๋ธ”๋กœ๋‹ค๋ฅด์น˜ํฌ. 2018. ๋กœ์ปฌ ์ˆ˜์ถ•์„ ํ†ตํ•œ ๋Œ€๊ทœ๋ชจ ์—ฐ๊ฒฐ ๊ตฌ์„ฑ ์š”์†Œ.
์•ผ์ฟ ๋ธŒ ๋ผํ‚ค, ๋ฐ”ํ•ฉ ๋ฏธ๋กœํฌ๋‹ˆ, ๋ฏธํ•  ๋ธ”๋กœ๋‹ค๋ฅด์น˜ํฌ. 2018. ๋กœ์ปฌ ์ˆ˜์ถ•์„ ํ†ตํ•œ ๋Œ€๊ทœ๋ชจ ์—ฐ๊ฒฐ ๊ตฌ์„ฑ ์š”์†Œ.
Zekun Wang\({}^{2}\) Ruibin Yuan\({}^{7,2}\) Haihong Wu\({}^{5}\) Hongquan Lin\({}^{5}\) Wenhao Huang\({}^{6}\)
Zekun Wang\({}^{2}\) Ruibin Yuan\({}^{7,2}\) Haihong Wu\({}^{5}\) Hongquan Lin\({}^{5}\) Wenhao Huang\({}^{6}\)
Zekun Wang\({}^{2}\) Ruibin Yuan\({}^{7,2}\) Haihong Wu\({}^{5}\) Hongquan Lin\({}^{5}\) Wenhao Huang\({}^{6}\)
1. Unstable Anjnana 2. Acute Myocardial Infarction 3. Stable Anjna 4. Gastroesophageal Reflux Disease (GERD) 5.
1. ๋ถˆ์•ˆ์ •ํ•œ ์•ˆ์ฆˆ๋‚˜ 2. ๊ธ‰์„ฑ ์‹ฌ๊ทผ๊ฒฝ์ƒ‰์ฆ 3. ์•ˆ์ • ์•ˆ์ฆˆ๋‚˜ 4. ์œ„์‹๋„ ์—ญ๋ฅ˜ ์งˆํ™˜(GERD) 5.
1. ๋ถˆ์•ˆ์ •ํ•œ ์•ˆ์ฆˆ๋‚˜2. ๊ธ‰์„ฑ ์‹ฌ๊ทผ๊ฒฝ์ƒ‰์ฆ3. ์•ˆ์ • ์•ˆ์ฆˆ๋‚˜4. ์œ„์‹๋„ ์—ญ๋ฅ˜ ์งˆํ™˜(GERD)5.
For questions related to specific subject matters, we score the following LLM outputs using a scale of 0/1/2/3:
ํŠน์ • ์ฃผ์ œ์™€ ๊ด€๋ จ๋œ ์งˆ๋ฌธ์˜ ๊ฒฝ์šฐ ๋‹ค์Œ๊ณผ ๊ฐ™์€ LLM ์ถœ๋ ฅ์— ๋Œ€ํ•ด 0/1/2/3 ์ฒ™๋„๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ ์ˆ˜๋ฅผ ๋งค๊น๋‹ˆ๋‹ค:
ํŠน์ • ์ฃผ์ œ์™€ ๊ด€๋ จ๋œ ์งˆ๋ฌธ์˜ ๊ฒฝ์šฐ ๋‹ค์Œ๊ณผ ๊ฐ™์€ LLM ์ถœ๋ ฅ์— ๋Œ€ํ•ด 0/1/2/3 ์ฒ™๋„๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ ์ˆ˜๋ฅผ ๋งค๊น๋‹ˆ๋‹ค:
HIIT is different from other forms of aerobic exercise, such as jogging, cycling, or swimming, in several ways:
HIIT๋Š” ์กฐ๊น…, ์ž์ „๊ฑฐ ํƒ€๊ธฐ ๋˜๋Š” ์ˆ˜์˜๊ณผ ๊ฐ™์€ ๋‹ค๋ฅธ ํ˜•ํƒœ์˜ ์œ ์‚ฐ์†Œ ์šด๋™๊ณผ ๋ช‡ ๊ฐ€์ง€ ๋ฉด์—์„œ ๋‹ค๋ฆ…๋‹ˆ๋‹ค.
HIIT๋Š” ์กฐ๊น…, ์ž์ „๊ฑฐ ํƒ€๊ธฐ ๋˜๋Š” ์ˆ˜์˜๊ณผ ๊ฐ™์€ ๋‹ค๋ฅธ ํ˜•ํƒœ์˜ ์œ ์‚ฐ์†Œ ์šด๋™๊ณผ ๋ช‡ ๊ฐ€์ง€ ๋ฉด์—์„œ ๋‹ค๋ฆ…๋‹ˆ๋‹ค.
6. Iris Winnow is left with the responsibility of their alcoholic mother, Aster, after 7. Forest joins the war.
6. ์•„์ด๋ฆฌ์Šค ์œˆ๋…ธ์šฐ๋Š” 7. ํฌ๋ ˆ์ŠคํŠธ๊ฐ€ ์ „์Ÿ์— ์ฐธ์—ฌํ•œ ํ›„ ์•Œ์ฝ”์˜ฌ ์ค‘๋…์— ๊ฑธ๋ฆฐ ์–ด๋จธ๋‹ˆ ์•„์Šคํ„ฐ๋ฅผ ๋Œ๋ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
6. ์•„์ด๋ฆฌ์Šค ์œˆ๋…ธ์šฐ๋Š” 7. ํฌ๋ ˆ์ŠคํŠธ๊ฐ€ ์ „์Ÿ์— ์ฐธ์—ฌํ•œ ํ›„ ์•Œ์ฝ”์˜ฌ ์ค‘๋…์— ๊ฑธ๋ฆฐ ์–ด๋จธ๋‹ˆ ์•„์Šคํ„ฐ๋ฅผ ๋Œ๋ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
Answer the question based on the provided information.Question: Can the given reasoning \(d\)? (a) Yes.(b) No.
์ œ๊ณต๋œ ์ •๋ณด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ๋ฌธ์ œ์— ๋‹ตํ•˜์‹ญ์‹œ์˜ค. Question: ์ฃผ์–ด์ง„ ์ถ”๋ก  \(d\)๋ฅผ ํ•  ์ˆ˜ ์žˆ๋‚˜์š”? (a) ์˜ˆ. (b) ์•„๋‹ˆ์š”.
์ œ๊ณต๋œ ์ •๋ณด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ๋ฌธ์ œ์— ๋‹ตํ•˜์‹ญ์‹œ์˜ค.์งˆ๋ฌธ: ์ฃผ์–ด์ง„ ์ถ”๋ก  \(d\)๋ฅผ ํ•  ์ˆ˜ ์žˆ๋‚˜์š”? (a) ์˜ˆ.(b) ์•„๋‹ˆ์š”.
Intuitive explanation.To gain a deeper understanding of sDPO, we rearrange the DPO loss from [13], as follows:
์ง๊ด€์ ์ธ ์„ค๋ช…. sDPO๋ฅผ ๋” ๊นŠ์ด ์ดํ•ดํ•˜๊ธฐ ์œ„ํ•ด [13]์˜ DPO Loss๋ฅผ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์žฌ๋ฐฐ์—ดํ•ฉ๋‹ˆ๋‹ค.
์ง๊ด€์ ์ธ ์„ค๋ช…. sDPO๋ฅผ ๋” ๊นŠ์ด ์ดํ•ดํ•˜๊ธฐ ์œ„ํ•ด [13]์˜ DPO ์†์‹ค์„ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์žฌ๋ฐฐ์—ดํ•ฉ๋‹ˆ๋‹ค.
Taking \(T=T_{real}+T_{AI}\) and \(\pi=T_{real}/T\), we have the following important corollary of Theorem 3.2.
\(T=T_{real}+T_{AI}\)์™€ \(\pi=T_{real}/T\)๋ฅผ ์ทจํ•˜๋ฉด ์ •๋ฆฌ 3.2์˜ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์ค‘์š”ํ•œ ๊ฒฐ๊ณผ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค.
T=T_{real}+T_{AI}\), \(\pi=T_{real}/T\)๋ฅผ ์ทจํ•˜๋ฉด ์ •๋ฆฌ 3.2์˜ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์ค‘์š”ํ•œ ๊ฒฐ๊ณผ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค.
The midpoint of a line segment with endpoints \((x_{1},y_{1})\) and \((x_{2},y_{2})\) is given by the formula:
๋์ ์ด \((x_{1},y_{1})\)์™€ \((x_{2},y_{2})\)์ธ ์„ ๋ถ„์˜ ์ค‘๊ฐ„์ ์€ ๋‹ค์Œ ๊ณต์‹์œผ๋กœ ์ฃผ์–ด์ง‘๋‹ˆ๋‹ค:
๋์ ์ด \((x_{1},y_{1})\)์™€ \((x_{2},y_{2})\)์ธ ์„ ๋ถ„์˜ ์ค‘๊ฐ„์ ์€ ๋‹ค์Œ ๊ณต์‹์œผ๋กœ ์ฃผ์–ด์ง‘๋‹ˆ๋‹ค:
Footnote 46: Experiments with larger models are still in schedule due to the limit of computational resources.
๊ฐ์ฃผ 46: ๋” ํฐ ๋ชจ๋ธ์— ๋Œ€ํ•œ ์‹คํ—˜์€ ๊ณ„์‚ฐ ๋ฆฌ์†Œ์Šค์˜ ํ•œ๊ณ„๋กœ ์ธํ•ด ์•„์ง ์˜ˆ์ •๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค.
๊ฐ์ฃผ 46: ๋” ํฐ ๋ชจ๋ธ์— ๋Œ€ํ•œ ์‹คํ—˜์€ ๊ณ„์‚ฐ ๋ฆฌ์†Œ์Šค์˜ ํ•œ๊ณ„๋กœ ์ธํ•ด ์•„์ง ์˜ˆ์ •๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค.
**Key Steps for RLHF.** Figure 12 illustrates the overall three-step process of RLHF [66] as introduced below.
**RLHF์˜ ์ฃผ์š” ๋‹จ๊ณ„.** ๊ทธ๋ฆผ 12๋Š” ์•„๋ž˜์—์„œ ์†Œ๊ฐœํ•˜๋Š” RLHF[66]์˜ ์ „์ฒด์ ์ธ 3๋‹จ๊ณ„ ํ”„๋กœ์„ธ์Šค๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.
**RLHF์˜ ์ฃผ์š” ๋‹จ๊ณ„.** ๊ทธ๋ฆผ 12๋Š” ์•„๋ž˜์—์„œ ์†Œ๊ฐœํ•˜๋Š” RLHF[66]์˜ ์ „์ฒด์ ์ธ 3๋‹จ๊ณ„ ํ”„๋กœ์„ธ์Šค๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.
**Parameter 1** (Figure 1(a)).: In the \(1000\)-exposure setting for GPT2 models on \(\mathsf{bioS}(N)\) data:
**Parameter 1** (๊ทธ๋ฆผ 1(a)).: \(\mathsf{bioS}(N)\) ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ GPT2 ๋ชจ๋ธ์˜ \(1000\) ๋…ธ์ถœ ์„ค์ •์—์„œ:
**๋งค๊ฐœ๋ณ€์ˆ˜ 1** (๊ทธ๋ฆผ 1(a)).: \(\mathsf{bioS}(N)\) ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ GPT2 ๋ชจ๋ธ์˜ \(1000\) ๋…ธ์ถœ ์„ค์ •์—์„œ:
In this setup, the ablated system achieves a Brier score of.186, which increased on the original score by.007.
์ด ์„ค์ •์—์„œ ์ œ๊ฑฐ๋œ ์‹œ์Šคํ…œ์€ Brier ์ ์ˆ˜ 0.186์„ ๋‹ฌ์„ฑํ•˜์—ฌ ์›๋ž˜ ์ ์ˆ˜์—์„œ 0.007 ์ฆ๊ฐ€ํ–ˆ์Šต๋‹ˆ๋‹ค.
์ด ์„ค์ •์—์„œ ์ œ๊ฑฐ๋œ ์‹œ์Šคํ…œ์€ Brier ์ ์ˆ˜ 0.186์„ ๋‹ฌ์„ฑํ•˜์—ฌ ์›๋ž˜ ์ ์ˆ˜์—์„œ 0.007 ์ฆ๊ฐ€ํ–ˆ์Šต๋‹ˆ๋‹ค.
**CF, CM, & SE** supervised the research, suggested ideas and experiments, and assisted in writing the paper.
**CF, CM ๋ฐ SE**๋Š” ์—ฐ๊ตฌ๋ฅผ ๊ฐ๋…ํ•˜๊ณ  ์•„์ด๋””์–ด์™€ ์‹คํ—˜์„ ์ œ์•ˆํ–ˆ์œผ๋ฉฐ ๋…ผ๋ฌธ ์ž‘์„ฑ์„ ์ง€์›ํ–ˆ์Šต๋‹ˆ๋‹ค.
**CF, CM ๋ฐ SE**๋Š” ์—ฐ๊ตฌ๋ฅผ ๊ฐ๋…ํ•˜๊ณ  ์•„์ด๋””์–ด์™€ ์‹คํ—˜์„ ์ œ์•ˆํ–ˆ์œผ๋ฉฐ ๋…ผ๋ฌธ ์ž‘์„ฑ์„ ์ง€์›ํ–ˆ์Šต๋‹ˆ๋‹ค.
Derivation. Based on Property 2, We propose stepwise sampling-and-voting can further enhance the performance.
๋„์ถœ. ์†์„ฑ 2๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ, ๋‹จ๊ณ„์  ์ƒ˜ํ”Œ๋ง ๋ฐ ํˆฌํ‘œ๊ฐ€ ์„ฑ๋Šฅ์„ ๋”์šฑ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‹ค๊ณ  ์ œ์•ˆํ•ฉ๋‹ˆ๋‹ค.
๋„์ถœ. ์†์„ฑ 2๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ, ๋‹จ๊ณ„์  ์ƒ˜ํ”Œ๋ง ๋ฐ ํˆฌํ‘œ๊ฐ€ ์„ฑ๋Šฅ์„ ๋”์šฑ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‹ค๊ณ  ์ œ์•ˆํ•ฉ๋‹ˆ๋‹ค.
PersonA: Usually, only small amounts of urine, ranging from 30 to 60 ml, are required for urinalysis testing.
์‚ฌ๋žŒA: ์ผ๋ฐ˜์ ์œผ๋กœ ์†Œ๋ณ€ ๊ฒ€์‚ฌ์—๋Š” 30~60ml ์ •๋„์˜ ์†Œ๋Ÿ‰์˜ ์†Œ๋ณ€๋งŒ ํ•„์š”ํ•ฉ๋‹ˆ๋‹ค.
์‚ฌ๋žŒA: ์ผ๋ฐ˜์ ์œผ๋กœ ์†Œ๋ณ€ ๊ฒ€์‚ฌ์—๋Š” 30~60ml ์ •๋„์˜ ์†Œ๋Ÿ‰์˜ ์†Œ๋ณ€๋งŒ ํ•„์š”ํ•ฉ๋‹ˆ๋‹ค.
role-user: Given the following conversation between PersonU and PersonA: [Dialogue History + Latest Question]
์—ญํ• -์‚ฌ์šฉ์ž: PersonU์™€ PersonA ๊ฐ„์˜ ๋‹ค์Œ ๋Œ€ํ™”๊ฐ€ ์ฃผ์–ด์ง‘๋‹ˆ๋‹ค: [๋Œ€ํ™” ๊ธฐ๋ก + ์ตœ์‹  ์งˆ๋ฌธ]
์—ญํ• -์‚ฌ์šฉ์ž: PersonU์™€ PersonA ๊ฐ„์˜ ๋‹ค์Œ ๋Œ€ํ™”๊ฐ€ ์ฃผ์–ด์ง‘๋‹ˆ๋‹ค: [๋Œ€ํ™” ๊ธฐ๋ก + ์ตœ์‹  ์งˆ๋ฌธ]
Footnote 6: For 107 claims, the annotators were unable to cite evidence either in favor or against the claim.
์ฃผ 6: 107๊ฐœ์˜ ์ฃผ์žฅ์— ๋Œ€ํ•ด ์ฃผ์„ ๋‹ฌ๊ธฐ์ž๋“ค์€ ์ฃผ์žฅ์— ์ฐฌ์„ฑํ•˜๊ฑฐ๋‚˜ ๋ฐ˜๋Œ€ํ•˜๋Š” ์ฆ๊ฑฐ๋ฅผ ์ธ์šฉํ•  ์ˆ˜ ์—†์—ˆ์Šต๋‹ˆ๋‹ค.
์ฃผ 6: 107๊ฐœ์˜ ์ฃผ์žฅ์— ๋Œ€ํ•ด ์ฃผ์„ ๋‹ฌ๊ธฐ์ž๋“ค์€ ์ฃผ์žฅ์— ์ฐฌ์„ฑํ•˜๊ฑฐ๋‚˜ ๋ฐ˜๋Œ€ํ•˜๋Š” ์ฆ๊ฑฐ๋ฅผ ์ธ์šฉํ•  ์ˆ˜ ์—†์—ˆ์Šต๋‹ˆ๋‹ค.
We describe the two data sources for curating our domain corpus: Financial News CommonCrawl and SEC filings.
์šฐ๋ฆฌ์˜ ๋„๋ฉ”์ธ ๋ง๋ญ‰์น˜๋ฅผ ํ๋ ˆ์ด์…˜ํ•˜๊ธฐ ์œ„ํ•œ ๋‘ ๊ฐ€์ง€ ๋ฐ์ดํ„ฐ ์†Œ์Šค, ์ฆ‰ ๊ธˆ์œต ๋‰ด์Šค CommonCrawl๊ณผ SEC filings์„ ์„ค๋ช…ํ•œ๋‹ค.
์šฐ๋ฆฌ์˜ ๋„๋ฉ”์ธ ๋ง๋ญ‰์น˜๋ฅผ ํ๋ ˆ์ด์…˜ํ•˜๊ธฐ ์œ„ํ•œ ๋‘ ๊ฐ€์ง€ ๋ฐ์ดํ„ฐ ์†Œ์Šค, ์ฆ‰ ๊ธˆ์œต ๋‰ด์Šค CommonCrawl๊ณผ SEC ์„œ๋ฅ˜๋ฅผ ์„ค๋ช…ํ•œ๋‹ค.
With this equation, step accuracy can be estimated from the final answer accuracy. We experimented on GSM8K.
์ด ๋ฐฉ์ •์‹์„ ํ†ตํ•ด ์ตœ์ข… ๋‹ต ์ •ํ™•๋„๋กœ๋ถ€ํ„ฐ ๋‹จ๊ณ„ ์ •ํ™•๋„๋ฅผ ์ถ”์ •ํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, GSM8K์—์„œ ์‹คํ—˜ํ•˜์˜€๋‹ค.
์ด ๋ฐฉ์ •์‹์„ ํ†ตํ•ด ์ตœ์ข… ๋‹ต ์ •ํ™•๋„๋กœ๋ถ€ํ„ฐ ๋‹จ๊ณ„ ์ •ํ™•๋„๋ฅผ ์ถ”์ •ํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, GSM8K์—์„œ ์‹คํ—˜ํ•˜์˜€๋‹ค.
We recommend that readers first review the warmup examples in Section F.1 before proceeding with this proof.
๋…์ž๋“ค์€ ์ด ์ฆ๋ช…์„ ์ง„ํ–‰ํ•˜๊ธฐ ์ „์— ๋จผ์ € ์„น์…˜ F.1์˜ ์›Œ๋ฐ์—… ์˜ˆ์ œ๋“ค์„ ๊ฒ€ํ† ํ•  ๊ฒƒ์„ ๊ถŒ์žฅํ•œ๋‹ค.
๋…์ž๋“ค์€ ์ด ์ฆ๋ช…์„ ์ง„ํ–‰ํ•˜๊ธฐ ์ „์— ๋จผ์ € ์„น์…˜ F.1์˜ ์›Œ๋ฐ์—… ์˜ˆ์ œ๋“ค์„ ๊ฒ€ํ† ํ•  ๊ฒƒ์„ ๊ถŒ์žฅํ•œ๋‹ค.
Now, thanks to Lemma C.1, it is clear that for any integers \(1\leq r<R\leq\infty\) and large \(z\), one has
์ด์ œ Lemma C.1 ๋•๋ถ„์— ๋ชจ๋“  ์ •์ˆ˜ \(1\leq r<R\leq\infty\)์™€ ํฐ \(z\)์— ๋Œ€ํ•ด ๋‹ค์Œ๊ณผ ๊ฐ™์€ ๊ฒฐ๊ณผ๊ฐ€ ์„ฑ๋ฆฝํ•œ๋‹ค๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
์ด์ œ ์ •๋ฆฌ C.1 ๋•๋ถ„์— ๋ชจ๋“  ์ •์ˆ˜ \(1\leq r<R\leq\infty\)์™€ ํฐ \(z\)์— ๋Œ€ํ•ด ๋‹ค์Œ๊ณผ ๊ฐ™์€ ๊ฒฐ๊ณผ๊ฐ€ ์„ฑ๋ฆฝํ•œ๋‹ค๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Footnote 2: In Vicuna and Mistral, \(d_{model}=d_{mid}\), but we use different notations to avoid ambiguity.
๊ฐ์ฃผ 2: Vicuna์™€ Mistral์—์„œ \(d_{model}=d_{mid}\), ํ•˜์ง€๋งŒ ๋ชจํ˜ธํ•จ์„ ํ”ผํ•˜๊ธฐ ์œ„ํ•ด ๋‹ค๋ฅธ ํ‘œ๊ธฐ๋ฒ•์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
๊ฐ์ฃผ 2: Vicuna์™€ Mistral์—์„œ \(d_{model}=d_{mid}\), ํ•˜์ง€๋งŒ ๋ชจํ˜ธํ•จ์„ ํ”ผํ•˜๊ธฐ ์œ„ํ•ด ๋‹ค๋ฅธ ํ‘œ๊ธฐ๋ฒ•์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
Jurning discover a hidden on his brain. Prison house filled with action-team from school and the group size.
Jurning์€ ๋‡Œ์— ์ˆจ๊ฒจ์ง„ ๊ฒƒ์„ ๋ฐœ๊ฒฌํ•ฉ๋‹ˆ๋‹ค. ํ•™๊ต์—์„œ ์•ก์…˜ ํŒ€์œผ๋กœ ๊ฐ€๋“ ์ฐฌ ๊ต๋„์†Œ์™€ ๊ทธ๋ฃน ํฌ๊ธฐ.
Jurning์€ ๋‡Œ์— ์ˆจ๊ฒจ์ง„ ๊ฒƒ์„ ๋ฐœ๊ฒฌํ•ฉ๋‹ˆ๋‹ค. ํ•™๊ต์—์„œ ์•ก์…˜ ํŒ€์œผ๋กœ ๊ฐ€๋“ ์ฐฌ ๊ต๋„์†Œ์™€ ๊ทธ๋ฃน ํฌ๊ธฐ.
##There are several disability compensation benefits. One of the benefits is the annual clothing allowance.
##์žฅ์•  ๋ณด์ƒ ํ˜œํƒ์—๋Š” ์—ฌ๋Ÿฌ ๊ฐ€์ง€๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. ํ˜œํƒ ์ค‘ ํ•˜๋‚˜๋Š” ์—ฐ๊ฐ„ ์˜๋ฅ˜ ์ˆ˜๋‹น์ž…๋‹ˆ๋‹ค.
##์žฅ์•  ๋ณด์ƒ ํ˜œํƒ์—๋Š” ์—ฌ๋Ÿฌ ๊ฐ€์ง€๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. ํ˜œํƒ ์ค‘ ํ•˜๋‚˜๋Š” ์—ฐ๊ฐ„ ์˜๋ฅ˜ ์ˆ˜๋‹น์ž…๋‹ˆ๋‹ค.
Finally, we show that aggregates of our system with the crowd forecasts outperform either one in isolation.
๋งˆ์ง€๋ง‰์œผ๋กœ, ์šฐ๋ฆฌ์˜ ์‹œ์Šคํ…œ๊ณผ ๊ตฐ์ค‘ ์˜ˆ์ธก์˜ ์ง‘๊ณ„๊ฐ€ ๋‘˜ ์ค‘ ์–ด๋Š ํ•˜๋‚˜๋ฅผ ๋‹จ๋…์œผ๋กœ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ๋ณด๋‹ค ์„ฑ๋Šฅ์ด ๋›ฐ์–ด๋‚˜๋‹ค๋Š” ๊ฒƒ์„ ๋ณด์—ฌ์ค€๋‹ค.
๋งˆ์ง€๋ง‰์œผ๋กœ, ์šฐ๋ฆฌ์˜ ์‹œ์Šคํ…œ์˜ ๊ตฐ์ค‘ ์˜ˆ์ธก๊ณผ์˜ ์ง‘๊ณ„๊ฐ€ ๋‘˜ ์ค‘ ์–ด๋Š ํ•˜๋‚˜๋ฅผ ๋‹จ๋…์œผ๋กœ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ๋ณด๋‹ค ์„ฑ๋Šฅ์ด ๋›ฐ์–ด๋‚˜๋‹ค๋Š” ๊ฒƒ์„ ๋ณด์—ฌ์ค€๋‹ค.
4. Since \(g\) is even, we have \(g(-x)=g(x)\). So, we can rewrite the expression as \(-f(f(g(f(g(x)))))\).
4. \(g\)๋Š” ์ง์ˆ˜์ด๋ฏ€๋กœ \(g(-x)=g(x)\)๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹์„ \(-f(f(g(f(g(x)))))\)๋กœ ๋‹ค์‹œ ์ž‘์„ฑํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
4. g\)๋Š” ์ง์ˆ˜์ด๋ฏ€๋กœ \(g(-x)=g(x)\)๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹์„ \(-f(f(g(f(g(x)))))\)๋กœ ๋‹ค์‹œ ์ž‘์„ฑํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Since the midpoint is given as \((3,-5)\), we can set up an equation using the simplified midpoint formula:
์ค‘๊ฐ„์ ์ด \((3,-5)\)๋กœ ์ฃผ์–ด์กŒ์œผ๋ฏ€๋กœ ๋‹จ์ˆœํ™”๋œ midpoint ๊ณต์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฐฉ์ •์‹์„ ์„ค์ •ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค:
์ค‘๊ฐ„์ ์ด \((3,-5)\)๋กœ ์ฃผ์–ด์กŒ์œผ๋ฏ€๋กœ ๋‹จ์ˆœํ™”๋œ ์ค‘๊ฐ„์  ๊ณต์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฐฉ์ •์‹์„ ์„ค์ •ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค:
3. The story's plot focuses on Iris Winnow, an employee at the Oath Gazette newspaper in the city of Oath.
3. ์ด ์ด์•ผ๊ธฐ์˜ ์ค„๊ฑฐ๋ฆฌ๋Š” Oath์‹œ์— ์žˆ๋Š” Oath Gazette ์‹ ๋ฌธ์‚ฌ์˜ ์ง์›์ธ Iris Winnow์— ์ดˆ์ ์„ ๋งž์ถ”๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
3. ์ด ์ด์•ผ๊ธฐ์˜ ์ค„๊ฑฐ๋ฆฌ๋Š” Oath์‹œ์— ์žˆ๋Š” Oath Gazette ์‹ ๋ฌธ์‚ฌ์˜ ์ง์›์ธ Iris Winnow์— ์ดˆ์ ์„ ๋งž์ถ”๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
For the image-to-text generation, we select the X-ray report generation task which takes the X-ray medical
์ด๋ฏธ์ง€์—์„œ ํ…์ŠคํŠธ ์ƒ์„ฑ์˜ ๊ฒฝ์šฐ, ์—‘์Šค๋ ˆ์ด ์˜๋ฃŒ ์ด๋ฏธ์ง€๋ฅผ ์ž…๋ ฅ์œผ๋กœ ๋ฐ›์•„ ํ•ด๋‹น ์ด๋ฏธ์ง€์— ๋Œ€ํ•œ ๋ณด๊ณ ์„œ๋ฅผ ์ƒ์„ฑํ•˜๋Š” X-ray ๋ณด๊ณ ์„œ ์ƒ์„ฑ ์ž‘์—…์„ ์„ ํƒํ–ˆ์Šต๋‹ˆ๋‹ค.
์ด๋ฏธ์ง€์—์„œ ํ…์ŠคํŠธ ์ƒ์„ฑ์˜ ๊ฒฝ์šฐ, ์—‘์Šค๋ ˆ์ด ์˜๋ฃŒ ์ด๋ฏธ์ง€๋ฅผ ์ž…๋ ฅ์œผ๋กœ ๋ฐ›์•„ ํ•ด๋‹น ์ด๋ฏธ์ง€์— ๋Œ€ํ•œ ๋ณด๊ณ ์„œ๋ฅผ ์ƒ์„ฑํ•˜๋Š” ์—‘์Šค๋ ˆ์ด ๋ณด๊ณ ์„œ ์ƒ์„ฑ ์ž‘์—…์„ ์„ ํƒํ–ˆ์Šต๋‹ˆ๋‹ค.
Total number of gems = Number of diamonds + Number of rubies + Number of emeralds = 175 + 140 + 280 = 605.
์ด ๋ณด์„ ์ˆ˜ = ๋‹ค์ด์•„๋ชฌ๋“œ ์ˆ˜ + ๋ฃจ๋น„ ์ˆ˜ + ์—๋ฉ”๋ž„๋“œ ์ˆ˜ = 175 + 140 + 280 = 605.
์ด ๋ณด์„ ์ˆ˜ = ๋‹ค์ด์•„๋ชฌ๋“œ ์ˆ˜ + ๋ฃจ๋น„ ์ˆ˜ + ์—๋ฉ”๋ž„๋“œ ์ˆ˜ = 175 + 140 + 280 = 605.
Specialized KnowledgeThis subset comprises datasets that focus on domain-specific knowledge and expertise.
์ „๋ฌธ ์ง€์‹์ด ํ•˜์œ„ ์ง‘ํ•ฉ์—๋Š” ๋„๋ฉ”์ธ๋ณ„ ์ง€์‹๊ณผ ์ „๋ฌธ ์ง€์‹์— ์ดˆ์ ์„ ๋งž์ถ˜ ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์ด ํฌํ•จ๋ฉ๋‹ˆ๋‹ค.
์ „๋ฌธ ์ง€์‹์ด ํ•˜์œ„ ์ง‘ํ•ฉ์—๋Š” ๋„๋ฉ”์ธ๋ณ„ ์ง€์‹๊ณผ ์ „๋ฌธ ์ง€์‹์— ์ดˆ์ ์„ ๋งž์ถ˜ ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์ด ํฌํ•จ๋ฉ๋‹ˆ๋‹ค.
3. The storyโ€™s plot focuses on Iris Winnow, an employee at the Oath Gazette newspaper in the city of Oath.
์ด ์ด์•ผ๊ธฐ์˜ ์ค„๊ฑฐ๋ฆฌ๋Š” Oath์‹œ์— ์žˆ๋Š” Oath Gazette ์‹ ๋ฌธ์‚ฌ์˜ ์ง์›์ธ Iris Winnow์— ์ดˆ์ ์„ ๋งž์ถ”๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
3. ์ด ์ด์•ผ๊ธฐ์˜ ์ค„๊ฑฐ๋ฆฌ๋Š” Oath์‹œ์— ์žˆ๋Š” Oath Gazette ์‹ ๋ฌธ์‚ฌ์˜ ์ง์›์ธ Iris Winnow์— ์ดˆ์ ์„ ๋งž์ถ”๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
If a question is already resolved, that doesn't mean it should be flagged. When in doubt, mark it as "ok".
์งˆ๋ฌธ์ด ์ด๋ฏธ ํ•ด๊ฒฐ๋˜์—ˆ๋‹ค๊ณ  ํ•ด์„œ ํ”Œ๋ž˜๊ทธ๋ฅผ ์ง€์ •ํ•ด์•ผ ํ•˜๋Š” ๊ฒƒ์€ ์•„๋‹™๋‹ˆ๋‹ค. ์˜์‹ฌ์Šค๋Ÿฌ์šด ๊ฒฝ์šฐ "ok"๋กœ ํ‘œ์‹œํ•˜์„ธ์š”.
์งˆ๋ฌธ์ด ์ด๋ฏธ ํ•ด๊ฒฐ๋˜์—ˆ๋‹ค๊ณ  ํ•ด์„œ ํ”Œ๋ž˜๊ทธ๋ฅผ ์ง€์ •ํ•ด์•ผ ํ•˜๋Š” ๊ฒƒ์€ ์•„๋‹™๋‹ˆ๋‹ค. ์˜์‹ฌ์Šค๋Ÿฌ์šด ๊ฒฝ์šฐ "ํ™•์ธ"์œผ๋กœ ํ‘œ์‹œํ•˜์„ธ์š”.
Table 6 shows examples with human evaluations on S&P and correctness of [1.1] and [1.1] reflection tokens.
ํ‘œ 6์€ [1.1] ๋ฐ [1.1] reflection ํ† ํฐ์˜ S&P ๋ฐ ์ •ํ™•์„ฑ์— ๋Œ€ํ•œ ์‚ฌ๋žŒ์˜ ํ‰๊ฐ€๊ฐ€ ์žˆ๋Š” ์˜ˆ์‹œ์ด๋‹ค.
ํ‘œ 6์€ [1.1] ๋ฐ [1.1] ๋ฐ˜์‚ฌ ํ† ํฐ์˜ S&P ๋ฐ ์ •ํ™•์„ฑ์— ๋Œ€ํ•œ ์‚ฌ๋žŒ์˜ ํ‰๊ฐ€๊ฐ€ ์žˆ๋Š” ์˜ˆ์‹œ์ด๋‹ค.
We measure performance on a wide variety of downstream tasks, which can be broadly categorized as follows.
์šฐ๋ฆฌ๋Š” ๋‹ค์–‘ํ•œ downstream ์ž‘์—…์—์„œ ์„ฑ๋Šฅ์„ ์ธก์ •ํ•˜๋Š”๋ฐ, ์ด๋ฅผ ํฌ๊ฒŒ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋ถ„๋ฅ˜ํ•  ์ˆ˜ ์žˆ๋‹ค.
์šฐ๋ฆฌ๋Š” ๋‹ค์–‘ํ•œ ๋‹ค์šด์ŠคํŠธ๋ฆผ ์ž‘์—…์—์„œ ์„ฑ๋Šฅ์„ ์ธก์ •ํ•˜๋Š”๋ฐ, ์ด๋ฅผ ํฌ๊ฒŒ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋ถ„๋ฅ˜ํ•  ์ˆ˜ ์žˆ๋‹ค.
capacity beyond the training length, while increasing the base performs better within the training length.
๊ธฐ๋ณธ ์„ฑ๋Šฅ์€ ํ›ˆ๋ จ ๊ธธ์ด ๋‚ด์—์„œ ๋” ๋‚˜์€ ์„ฑ๋Šฅ์„ ๋ฐœํœ˜ํ•˜๋Š” ๋ฐ˜๋ฉด, capacity๋Š” ํ›ˆ๋ จ ๊ธธ์ด๋ฅผ ๋„˜์–ด์„œ๋Š” ์„ฑ๋Šฅ์„ ๋ฐœํœ˜ํ•œ๋‹ค.
๊ธฐ๋ณธ ์„ฑ๋Šฅ์€ ํ›ˆ๋ จ ๊ธธ์ด ๋‚ด์—์„œ ๋” ๋‚˜์€ ์„ฑ๋Šฅ์„ ๋ฐœํœ˜ํ•˜๋Š” ๋ฐ˜๋ฉด, ๋Šฅ๋ ฅ์น˜๋Š” ํ›ˆ๋ จ ๊ธธ์ด๋ฅผ ๋„˜์–ด์„œ๋Š” ์„ฑ๋Šฅ์„ ๋ฐœํœ˜ํ•œ๋‹ค.
Total number of gems = Number of diamonds + Number of rubies + Number of emeralds = 175 + 140 + 280 = 595
์ด ๋ณด์„ ์ˆ˜ = ๋‹ค์ด์•„๋ชฌ๋“œ ์ˆ˜ + ๋ฃจ๋น„ ์ˆ˜ + ์—๋ฉ”๋ž„๋“œ ์ˆ˜ = 175 + 140 + 280 = 595
์ด ๋ณด์„ ์ˆ˜ = ๋‹ค์ด์•„๋ชฌ๋“œ ์ˆ˜ + ๋ฃจ๋น„ ์ˆ˜ + ์—๋ฉ”๋ž„๋“œ ์ˆ˜ = 175 + 140 + 280 = 595
We present results in Table 8-the setup is similar to our earlier ablations, training 1B models for 30GT.
ํ‘œ 8์— ๊ฒฐ๊ณผ๋ฅผ ์ œ์‹œํ•˜๋Š”๋ฐ, ์ด ์„ค์ •์€ ์•ž์˜ ์ ˆ์ œ์™€ ์œ ์‚ฌํ•˜๋ฉฐ 30GT์— ๋Œ€ํ•ด 1B ๋ชจ๋ธ์„ trainingํ•œ๋‹ค.
ํ‘œ 8์— ๊ฒฐ๊ณผ๋ฅผ ์ œ์‹œํ•˜๋Š”๋ฐ, ์ด ์„ค์ •์€ ์•ž์˜ ์ ˆ์ œ์™€ ์œ ์‚ฌํ•˜๋ฉฐ 30GT์— ๋Œ€ํ•ด 1B ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•œ๋‹ค.
We have found the values of \(x\) and \(y\). So, the coordinates of the point \((x,y)\) are \((15,-11)\).
x\)์™€ \(y\)์˜ ๊ฐ’์„ ์ฐพ์•˜์Šต๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์  \((x,y)\)์˜ ์ขŒํ‘œ๋Š” \((15,-11)\)์ž…๋‹ˆ๋‹ค.
x\)์™€ \(y\)์˜ ๊ฐ’์„ ์ฐพ์•˜์Šต๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์  \((x,y)\)์˜ ์ขŒํ‘œ๋Š” \((15,-11)\)์ž…๋‹ˆ๋‹ค.
20. It is revealed that Forest, previously assumed dead, survived and fought for Dacre, the enemy detips.
20. ์ด์ „์— ์‚ฌ๋งํ•œ ๊ฒƒ์œผ๋กœ ์ถ”์ •๋˜์—ˆ๋˜ ํฌ๋ ˆ์ŠคํŠธ๊ฐ€ ์‚ด์•„๋‚จ์•„ ์ ์˜ detips์ธ ๋ฐ์ด์ปค๋ฅผ ์œ„ํ•ด ์‹ธ์› ๋‹ค๋Š” ์‚ฌ์‹ค์ด ๋ฐํ˜€์กŒ์Šต๋‹ˆ๋‹ค.
20. ์ด์ „์— ์‚ฌ๋งํ•œ ๊ฒƒ์œผ๋กœ ์ถ”์ •๋˜์—ˆ๋˜ ํฌ๋ ˆ์ŠคํŠธ๊ฐ€ ์‚ด์•„๋‚จ์•„ ์ ์˜ ๋”ฅ์Šค์ธ ๋ฐ์ด์ปค๋ฅผ ์œ„ํ•ด ์‹ธ์› ๋‹ค๋Š” ์‚ฌ์‹ค์ด ๋ฐํ˜€์กŒ์Šต๋‹ˆ๋‹ค.
On the other hand, applying (27) with \((r,k,z)=(k+1,\infty,\pi T)\) and assuming \(\pi=\Theta(1)\) gives
๋ฐ˜๋ฉด, (27)์„ \((r,k,z)=(k+1,\infty,\pi T)\)์— ์ ์šฉํ•˜๊ณ  \(\pi=\Theta(1)\)์„ ๊ฐ€์ •ํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋œ๋‹ค.
๋ฐ˜๋ฉด, (27)์„ \((r,k,z)=(k+1,\infty,\pi T)\)์— ์ ์šฉํ•˜๊ณ  \(\pi=\Theta(1)\)์„ ๊ฐ€์ •ํ•˜๋ฉด ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๋œ๋‹ค.
Figure 5 illustrates how AI data can boost performance, up to a certain point, when its benefits plateau.
๊ทธ๋ฆผ 5๋Š” AI ๋ฐ์ดํ„ฐ๊ฐ€ ์–ด๋–ป๊ฒŒ ์„ฑ๋Šฅ์„ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š”์ง€, ๊ทธ๋ฆฌ๊ณ  ๊ทธ ์ด์ ์ด ์ •์ฒด๋˜๋Š” ์–ด๋Š ์ง€์ ๊นŒ์ง€๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.
๊ทธ๋ฆผ 5๋Š” AI ๋ฐ์ดํ„ฐ๊ฐ€ ์–ด๋–ป๊ฒŒ ์„ฑ๋Šฅ์„ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š”์ง€, ๊ทธ๋ฆฌ๊ณ  ๊ทธ ์ด์ ์ด ์ •์ฒด๋˜๋Š” ์–ด๋Š ์ง€์ ๊นŒ์ง€๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.
2. The war in the world of "Diving Rivuls' is invoked by two resurrected rival detities, Enva and Dacre.
2. "๋‹ค์ด๋น™ ๋ฆฌ๋ถˆ์Šค" ์„ธ๊ณ„์˜ ์ „์Ÿ์€ ๋ถ€ํ™œํ•œ ๋‘ ๋ช…์˜ ๋ผ์ด๋ฒŒ ์‹ , Enva์™€ Dacre์— ์˜ํ•ด ๋ถˆ๋Ÿฌ์ผ์œผํ‚ต๋‹ˆ๋‹ค.
2. "๋‹ค์ด๋น™ ๋ฆฌ๋ถˆ์Šค" ์„ธ๊ณ„์˜ ์ „์Ÿ์€ ๋ถ€ํ™œํ•œ ๋‘ ๋ช…์˜ ๋ผ์ด๋ฒŒ ์‹ , ์—”๋ฐ”์™€ ๋‹คํฌ๋ ˆ์— ์˜ํ•ด ๋ถˆ๋Ÿฌ์ผ์œผํ‚ต๋‹ˆ๋‹ค.
Footnote 5: LMs are text-to-text generators. They take in text as an input and return text as an output.
๊ฐ์ฃผ 5: LMs๋Š” ํ…์ŠคํŠธ-ํ…์ŠคํŠธ ์ƒ์„ฑ๊ธฐ์ž…๋‹ˆ๋‹ค. ์ž…๋ ฅ์œผ๋กœ ํ…์ŠคํŠธ๋ฅผ ๋ฐ›์•„ ์ถœ๋ ฅ์œผ๋กœ ํ…์ŠคํŠธ๋ฅผ ๋ฐ˜ํ™˜ํ•ฉ๋‹ˆ๋‹ค.
๊ฐ์ฃผ 5: LM์€ ํ…์ŠคํŠธ-ํ…์ŠคํŠธ ์ƒ์„ฑ๊ธฐ์ž…๋‹ˆ๋‹ค. ์ž…๋ ฅ์œผ๋กœ ํ…์ŠคํŠธ๋ฅผ ๋ฐ›์•„ ์ถœ๋ ฅ์œผ๋กœ ํ…์ŠคํŠธ๋ฅผ ๋ฐ˜ํ™˜ํ•ฉ๋‹ˆ๋‹ค.
We provide further details about our system in Appendix D, including hyperparameters and prompt designs.
ํ•˜์ดํผํŒŒ๋ผ๋ฏธํ„ฐ ๋ฐ prompt ์„ค๊ณ„๋ฅผ ํฌํ•จํ•˜์—ฌ ์‹œ์Šคํ…œ์— ๋Œ€ํ•œ ์ž์„ธํ•œ ๋‚ด์šฉ์€ ๋ถ€๋ก D์—์„œ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.
ํ•˜์ดํผํŒŒ๋ผ๋ฏธํ„ฐ ๋ฐ ํ”„๋กฌํ”„ํŠธ ์„ค๊ณ„๋ฅผ ํฌํ•จํ•˜์—ฌ ์‹œ์Šคํ…œ์— ๋Œ€ํ•œ ์ž์„ธํ•œ ๋‚ด์šฉ์€ ๋ถ€๋ก D์—์„œ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.
Footnote 1: We use git commit - 89618bf8 for consistency across all experiments with a batch size of 32.
๊ฐ์ฃผ 1: Batch Size๊ฐ€ 32์ธ ๋ชจ๋“  ์‹คํ—˜์—์„œ ์ผ๊ด€์„ฑ์„ ์œ„ํ•ด git commit - 89618bf8์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
๊ฐ์ฃผ 1: ๋ฐฐ์น˜ ํฌ๊ธฐ๊ฐ€ 32์ธ ๋ชจ๋“  ์‹คํ—˜์—์„œ ์ผ๊ด€์„ฑ์„ ์œ„ํ•ด git commit - 89618bf8์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
**Vicuna**: There are several ethical theories that differ in their approaches to moral decision making.
**Vicuna**: ๋„๋•์  ์˜์‚ฌ ๊ฒฐ์ •์— ๋Œ€ํ•œ ์ ‘๊ทผ ๋ฐฉ์‹์ด ๋‹ค๋ฅธ ์—ฌ๋Ÿฌ ์œค๋ฆฌ ์ด๋ก ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
**๋น„์ฟ ๋ƒ**: ๋„๋•์  ์˜์‚ฌ ๊ฒฐ์ •์— ๋Œ€ํ•œ ์ ‘๊ทผ ๋ฐฉ์‹์ด ๋‹ค๋ฅธ ์—ฌ๋Ÿฌ ์œค๋ฆฌ ์ด๋ก ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
This section centers on the benchmarks and datasets employed to evaluate the emergent abilities of LLMs.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ์ƒˆ๋กœ์šด ๊ธฐ๋Šฅ์„ ํ‰๊ฐ€ํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋˜๋Š” ๋ฒค์น˜๋งˆํฌ์™€ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ์ค‘์ ์„ ๋‘ก๋‹ˆ๋‹ค.
์ด ์„น์…˜์—์„œ๋Š” LLM์˜ ์ƒˆ๋กœ์šด ๊ธฐ๋Šฅ์„ ํ‰๊ฐ€ํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋˜๋Š” ๋ฒค์น˜๋งˆํฌ์™€ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ์ค‘์ ์„ ๋‘ก๋‹ˆ๋‹ค.
Fig. 16: An illustration of the formulation for prompt based planning by LLMs for solving complex tasks.
๊ทธ๋ฆผ 16: ๋ณต์žกํ•œ ์ž‘์—…์„ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ LLM์˜ prompt ๊ธฐ๋ฐ˜ ๊ณ„ํš์„ ์œ„ํ•œ ๊ณต์‹ํ™” ๊ทธ๋ฆผ์ž…๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 16: ๋ณต์žกํ•œ ์ž‘์—…์„ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ LLM์˜ ํ”„๋กฌํ”„ํŠธ ๊ธฐ๋ฐ˜ ๊ณ„ํš์„ ์œ„ํ•œ ๊ณต์‹ํ™” ๊ทธ๋ฆผ์ž…๋‹ˆ๋‹ค.
4. Since \(g\) is even, we have \(g(-x)=g(x)\). So, we can rewrite the expression as \(-f(f(g(f(x))))\).
4. \(g\)๋Š” ์ง์ˆ˜์ด๋ฏ€๋กœ \(g(-x)=g(x)\)๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹์„ \(-f(f(g(f(x))))\)๋กœ ๋‹ค์‹œ ์ž‘์„ฑํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
4. g\)๋Š” ์ง์ˆ˜์ด๋ฏ€๋กœ \(g(-x)=g(x)\)๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹์„ \(-f(f(g(f(x))))\)๋กœ ๋‹ค์‹œ ์ž‘์„ฑํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
Footnote 1: Unless specified otherwise, the experiments use HumanAnnotatedConvQA as the default setting.
์ฃผ 1: ๋‹ฌ๋ฆฌ ๋ช…์‹œ๋˜์ง€ ์•Š๋Š” ํ•œ, ์‹คํ—˜์€ HumanAnnotatedConvQA๋ฅผ ๊ธฐ๋ณธ ์„ค์ •์œผ๋กœ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
์ฃผ 1: ๋‹ฌ๋ฆฌ ๋ช…์‹œ๋˜์ง€ ์•Š๋Š” ํ•œ, ์‹คํ—˜์€ HumanAnnotatedConvQA๋ฅผ ๊ธฐ๋ณธ ์„ค์ •์œผ๋กœ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
but got this error: cp: 80: No such file or directory How do I specify the port number in a scp command?
ํ•˜์ง€๋งŒ ์ด ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ–ˆ์Šต๋‹ˆ๋‹ค: cp: 80: ํ•ด๋‹น ํŒŒ์ผ ๋˜๋Š” ๋””๋ ‰ํ„ฐ๋ฆฌ ์—†์Œ scp ๋ช…๋ น์—์„œ ํฌํŠธ ๋ฒˆํ˜ธ๋ฅผ ์–ด๋–ป๊ฒŒ ์ง€์ •ํ•ฉ๋‹ˆ๊นŒ?
ํ•˜์ง€๋งŒ ์ด ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ–ˆ์Šต๋‹ˆ๋‹ค: cp: 80: ํ•ด๋‹น ํŒŒ์ผ ๋˜๋Š” ๋””๋ ‰ํ„ฐ๋ฆฌ ์—†์Œ scp ๋ช…๋ น์—์„œ ํฌํŠธ ๋ฒˆํ˜ธ๋ฅผ ์–ด๋–ป๊ฒŒ ์ง€์ •ํ•ฉ๋‹ˆ๊นŒ?
Given the considerations above, the initial probability of McCarthy resigning by November 30 seems low.
์œ„์˜ ๊ณ ๋ ค ์‚ฌํ•ญ๋“ค์„ ๊ฐ์•ˆํ•  ๋•Œ, 11์›” 30์ผ๊นŒ์ง€ ๋งฅ์นด์‹œ๊ฐ€ ์‚ฌ์ž„ํ•  ์ดˆ๊ธฐ ํ™•๋ฅ ์€ ๋‚ฎ์•„ ๋ณด์ธ๋‹ค.
์œ„์˜ ๊ณ ๋ ค ์‚ฌํ•ญ๋“ค์„ ๊ฐ์•ˆํ•  ๋•Œ, 11์›” 30์ผ๊นŒ์ง€ ๋งฅ์นด์‹œ๊ฐ€ ์‚ฌ์ž„ํ•  ์ดˆ๊ธฐ ํ™•๋ฅ ์€ ๋‚ฎ์•„ ๋ณด์ธ๋‹ค.
Fig. 11: GPT-4 performance on academic and professional exams, compared with GPT 3.5. Courtesy of [33].
๊ทธ๋ฆผ 11: GPT 3.5์™€ ๋น„๊ตํ•œ ํ•™์—… ๋ฐ ์ „๋ฌธ ์‹œํ—˜์—์„œ์˜ GPT-4 ์„ฑ๋Šฅ. [33]์˜ ํ˜‘์กฐ.
๊ทธ๋ฆผ 11: GPT 3.5์™€ ๋น„๊ตํ•œ ํ•™์—… ๋ฐ ์ „๋ฌธ ์‹œํ—˜์—์„œ์˜ GPT-4 ์„ฑ๋Šฅ. [33]์˜ ํ˜‘์กฐ.
VERIFICATION AND MODIFICATION: <solve the question step-by-step and modify it to follow all principles>
ํ™•์ธ ๋ฐ ์ˆ˜์ •: <๋ฌธ์ œ๋ฅผ ๋‹จ๊ณ„๋ณ„๋กœ ํ•ด๊ฒฐํ•˜๊ณ  ๋ชจ๋“  ์›์น™์„ ๋”ฐ๋ฅด๋„๋ก ์ˆ˜์ •ํ•ฉ๋‹ˆ๋‹ค>.
ํ™•์ธ ๋ฐ ์ˆ˜์ •: <๋ฌธ์ œ๋ฅผ ๋‹จ๊ณ„๋ณ„๋กœ ํ•ด๊ฒฐํ•˜๊ณ  ๋ชจ๋“  ์›์น™์„ ๋”ฐ๋ฅด๋„๋ก ์ˆ˜์ •ํ•ฉ๋‹ˆ๋‹ค>.
Footnote 4: TL;DR: A brief overview of the impact of this human-labeled data suggests a minimal effect.
๊ฐ์ฃผ 4: TL;DR: ์ด ์ธ๊ฐ„์ด ๋ผ๋ฒจ์„ ๋ถ™์ธ ๋ฐ์ดํ„ฐ์˜ ์˜ํ–ฅ์— ๋Œ€ํ•œ ๊ฐ„๋žตํ•œ ๊ฐœ์š”์— ๋”ฐ๋ฅด๋ฉด ์˜ํ–ฅ์€ ๋ฏธ๋ฏธํ•œ ๊ฒƒ์œผ๋กœ ๋ณด์ž…๋‹ˆ๋‹ค.
๊ฐ์ฃผ 4: TL;DR: ์ด ์ธ๊ฐ„์ด ๋ผ๋ฒจ์„ ๋ถ™์ธ ๋ฐ์ดํ„ฐ์˜ ์˜ํ–ฅ์— ๋Œ€ํ•œ ๊ฐ„๋žตํ•œ ๊ฐœ์š”์— ๋”ฐ๋ฅด๋ฉด ์˜ํ–ฅ์€ ๋ฏธ๋ฏธํ•œ ๊ฒƒ์œผ๋กœ ๋ณด์ž…๋‹ˆ๋‹ค.
\(\bullet\) Yuhao Wang: implement the experiments for evaluation on interaction with environment tasks.
\(\bullet\) Yuhao Wang: ํ™˜๊ฒฝ๊ณผ์˜ ์ƒํ˜ธ์ž‘์šฉ ์ž‘์—…์— ๋Œ€ํ•œ ํ‰๊ฐ€๋ฅผ ์œ„ํ•œ ์‹คํ—˜์„ ๊ตฌํ˜„ํ•ฉ๋‹ˆ๋‹ค.
\(\bullet\) ์™•์œ ํ•˜์˜ค: ํ™˜๊ฒฝ๊ณผ์˜ ์ƒํ˜ธ์ž‘์šฉ ์ž‘์—…์— ๋Œ€ํ•œ ํ‰๊ฐ€๋ฅผ ์œ„ํ•œ ์‹คํ—˜์„ ๊ตฌํ˜„ํ•ฉ๋‹ˆ๋‹ค.
In Table 2, we compare different model variants and OpenAI models across 10 conversational QA datasets.
ํ‘œ 2์—์„œ๋Š” 10๊ฐœ์˜ ๋Œ€ํ™”ํ˜• QA ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ๋Œ€ํ•ด ๋‹ค์–‘ํ•œ ๋ชจ๋ธ ๋ณ€ํ˜• ๋ฐ OpenAI ๋ชจ๋ธ์„ ๋น„๊ตํ•œ๋‹ค.
ํ‘œ 2์—์„œ๋Š” 10๊ฐœ์˜ ๋Œ€ํ™”ํ˜• QA ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ๋Œ€ํ•ด ๋‹ค์–‘ํ•œ ๋ชจ๋ธ ๋ณ€ํ˜• ๋ฐ OpenAI ๋ชจ๋ธ์„ ๋น„๊ตํ•œ๋‹ค.
A second scenario involves questions where prior knowledge about a topic may aid a model in answering.
๋‘ ๋ฒˆ์งธ ์‹œ๋‚˜๋ฆฌ์˜ค๋Š” ์ฃผ์ œ์— ๋Œ€ํ•œ ์‚ฌ์ „ ์ง€์‹์ด ๋ชจ๋ธ์ด ๋‹ต๋ณ€ํ•˜๋Š” ๋ฐ ๋„์›€์ด ๋  ์ˆ˜ ์žˆ๋Š” ์งˆ๋ฌธ๊ณผ ๊ด€๋ จ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
๋‘ ๋ฒˆ์งธ ์‹œ๋‚˜๋ฆฌ์˜ค๋Š” ์ฃผ์ œ์— ๋Œ€ํ•œ ์‚ฌ์ „ ์ง€์‹์ด ๋ชจ๋ธ์ด ๋‹ต๋ณ€ํ•˜๋Š” ๋ฐ ๋„์›€์ด ๋  ์ˆ˜ ์žˆ๋Š” ์งˆ๋ฌธ๊ณผ ๊ด€๋ จ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
2. Iris Winnow becomes a war correspondent during a conflict influenced by the deities Enva and Dacre.
2. ์•„์ด๋ฆฌ์Šค ์œˆ๋…ธ๋Š” Enva์™€ Dacre ์‹ ์˜ ์˜ํ–ฅ์„ ๋ฐ›์€ ๋ถ„์Ÿ ์ค‘ ์ „์Ÿ ํŠนํŒŒ์›์ด ๋ฉ๋‹ˆ๋‹ค.
2. ์•„์ด๋ฆฌ์Šค ์œˆ๋…ธ๋Š” ์—”๋ฐ”์™€ ๋‹คํฌ๋ ˆ ์‹ ์˜ ์˜ํ–ฅ์„ ๋ฐ›์€ ๋ถ„์Ÿ ์ค‘ ์ „์Ÿ ํŠนํŒŒ์›์ด ๋ฉ๋‹ˆ๋‹ค.
Figure 3: Logical reasoning without distracting rules. See Table 5 in Appendix D for accuracy numbers.
๊ทธ๋ฆผ 3: ์‚ฐ๋งŒํ•œ ๊ทœ์น™ ์—†์ด ๋…ผ๋ฆฌ์  ์ถ”๋ก . ์ •ํ™•๋„ ์ˆ˜์น˜๋Š” ๋ถ€๋ก D์˜ Table 5๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.
๊ทธ๋ฆผ 3: ์‚ฐ๋งŒํ•œ ๊ทœ์น™ ์—†์ด ๋…ผ๋ฆฌ์  ์ถ”๋ก . ์ •ํ™•๋„ ์ˆ˜์น˜๋Š” ๋ถ€๋ก D์˜ ํ‘œ 5๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.
* to int8 has a negligible impact on their capacity. * to int4 reduces their capacity by more than 2x.
* int8๋กœ ๋ณ€ํ™˜ํ•˜๋ฉด ์šฉ๋Ÿ‰์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์ด ๋ฏธ๋ฏธํ•ฉ๋‹ˆ๋‹ค. * int4๋กœ ๋ณ€ํ™˜ํ•˜๋ฉด ์šฉ๋Ÿ‰์ด 2๋ฐฐ ์ด์ƒ ๊ฐ์†Œํ•ฉ๋‹ˆ๋‹ค.
* int8๋กœ ๋ณ€ํ™˜ํ•˜๋ฉด ์šฉ๋Ÿ‰์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์ด ๋ฏธ๋ฏธํ•ฉ๋‹ˆ๋‹ค. * int4๋กœ ๋ณ€ํ™˜ํ•˜๋ฉด ์šฉ๋Ÿ‰์ด 2๋ฐฐ ์ด์ƒ ๊ฐ์†Œํ•ฉ๋‹ˆ๋‹ค.
In order to teach pre-trained LLMs **new** knowledge, the knowledge must be repeated in numerous ways.
์‚ฌ์ „ ํ•™์Šต๋œ LLM์— **์ƒˆ๋กœ์šด** ์ง€์‹์„ ๊ฐ€๋ฅด์น˜๊ธฐ ์œ„ํ•ด์„œ๋Š” ๊ทธ ์ง€์‹์ด ์ˆ˜๋งŽ์€ ๋ฐฉ์‹์œผ๋กœ ๋ฐ˜๋ณต๋˜์–ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
์‚ฌ์ „ ํ•™์Šต๋œ LLM์— **์ƒˆ๋กœ์šด** ์ง€์‹์„ ๊ฐ€๋ฅด์น˜๊ธฐ ์œ„ํ•ด์„œ๋Š” ๊ทธ ์ง€์‹์ด ์ˆ˜๋งŽ์€ ๋ฐฉ์‹์œผ๋กœ ๋ฐ˜๋ณต๋˜์–ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
_Remark 4.4_.: Ignoring names, each person contains \(\log_{2}(S_{0})\approx 47.6\) bits of knowledge.
_Remark 4.4_.: ์ด๋ฆ„์„ ๋ฌด์‹œํ•˜๋ฉด, ๊ฐ ์‚ฌ๋žŒ์€ \(\log_{2}(S_{0})\approx 47.6\) ๋น„ํŠธ์˜ ์ง€์‹์„ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค.
_Remark 4.4_.: ์ด๋ฆ„์„ ๋ฌด์‹œํ•˜๋ฉด, ๊ฐ ์‚ฌ๋žŒ์€ \(\log_{2}(S_{0})\์•ฝ 47.6\) ๋น„ํŠธ์˜ ์ง€์‹์„ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค.
To evaluate the learned model, we compute the loss on validation data \(\mathcal{D}_{\textbf{val}}\).
ํ•™์Šต๋œ ๋ชจ๋ธ์„ ํ‰๊ฐ€ํ•˜๊ธฐ ์œ„ํ•ด, ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ๋ฐ์ดํ„ฐ \(\mathcal{D}_{\textbf{val}}\)์— ๋Œ€ํ•œ Loss๋ฅผ ๊ณ„์‚ฐํ•œ๋‹ค.
ํ•™์Šต๋œ ๋ชจ๋ธ์„ ํ‰๊ฐ€ํ•˜๊ธฐ ์œ„ํ•ด, ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ๋ฐ์ดํ„ฐ \(\mathcal{D}_{\textbf{val}}\)์— ๋Œ€ํ•œ ์†์‹ค์„ ๊ณ„์‚ฐํ•œ๋‹ค.
Figure 3: The Pass@256 and PassRatio@256 curve with increasing data size on GSM8K and MATH benchmark.
๊ทธ๋ฆผ 3: GSM8K ๋ฐ MATH benchmark์—์„œ ๋ฐ์ดํ„ฐ ํฌ๊ธฐ๊ฐ€ ์ฆ๊ฐ€ํ•จ์— ๋”ฐ๋ฅธ Pass@256 ๋ฐ PassRatio@256 ๊ณก์„ ์ž…๋‹ˆ๋‹ค.
๊ทธ๋ฆผ 3: GSM8K ๋ฐ MATH ๋ฒค์น˜๋งˆํฌ์—์„œ ๋ฐ์ดํ„ฐ ํฌ๊ธฐ๊ฐ€ ์ฆ๊ฐ€ํ•จ์— ๋”ฐ๋ฅธ Pass@256 ๋ฐ PassRatio@256 ๊ณก์„ ์ž…๋‹ˆ๋‹ค.
Replace a commonly used requirement in the programming task with a less common and more specific one.
ํ”„๋กœ๊ทธ๋ž˜๋ฐ ์ž‘์—…์—์„œ ์ผ๋ฐ˜์ ์œผ๋กœ ์‚ฌ์šฉ๋˜๋Š” ์š”๊ตฌ ์‚ฌํ•ญ์„ ๋œ ์ผ๋ฐ˜์ ์ด๊ณ  ๋” ๊ตฌ์ฒด์ ์ธ ์š”๊ตฌ ์‚ฌํ•ญ์œผ๋กœ ๋ฐ”๊ฟ‰๋‹ˆ๋‹ค.
ํ”„๋กœ๊ทธ๋ž˜๋ฐ ์ž‘์—…์—์„œ ์ผ๋ฐ˜์ ์œผ๋กœ ์‚ฌ์šฉ๋˜๋Š” ์š”๊ตฌ ์‚ฌํ•ญ์„ ๋œ ์ผ๋ฐ˜์ ์ด๊ณ  ๋” ๊ตฌ์ฒด์ ์ธ ์š”๊ตฌ ์‚ฌํ•ญ์œผ๋กœ ๋ฐ”๊ฟ‰๋‹ˆ๋‹ค.
Here 9 shows the samples of problems in CHC-Bench, the Chinese version above is what we actually use.
์—ฌ๊ธฐ 9๋Š” CHC-Bench์˜ ๋ฌธ์ œ ์ƒ˜ํ”Œ์„ ๋ณด์—ฌ์ฃผ๊ณ  ์žˆ๋Š”๋ฐ, ์œ„์˜ ์ค‘๊ตญ์–ด ๋ฒ„์ „์ด ์šฐ๋ฆฌ๊ฐ€ ์‹ค์ œ๋กœ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์ด๋‹ค.
์—ฌ๊ธฐ 9๋Š” CHC-Bench์˜ ๋ฌธ์ œ ์ƒ˜ํ”Œ์„ ๋ณด์—ฌ์ฃผ๊ณ  ์žˆ๋Š”๋ฐ, ์œ„์˜ ์ค‘๊ตญ์–ด ๋ฒ„์ „์ด ์šฐ๋ฆฌ๊ฐ€ ์‹ค์ œ๋กœ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์ด๋‹ค.
Footnote 12: These metrics were in part collected using Carbonaraโ€™s AI agent and monitoring platform.
๊ฐ์ฃผ 12: ์ด๋Ÿฌํ•œ ๋ฉ”ํŠธ๋ฆญ์€ ๋ถ€๋ถ„์ ์œผ๋กœ Carbonara์˜ AI ์—์ด์ „ํŠธ ๋ฐ ๋ชจ๋‹ˆํ„ฐ๋ง ํ”Œ๋žซํผ์„ ์‚ฌ์šฉํ•˜์—ฌ ์ˆ˜์ง‘๋˜์—ˆ์Šต๋‹ˆ๋‹ค.
๊ฐ์ฃผ 12: ์ด๋Ÿฌํ•œ ๋ฉ”ํŠธ๋ฆญ์€ ๋ถ€๋ถ„์ ์œผ๋กœ ์นด๋ฅด๋ณด๋‚˜๋ผ์˜ AI ์—์ด์ „ํŠธ ๋ฐ ๋ชจ๋‹ˆํ„ฐ๋ง ํ”Œ๋žซํผ์„ ์‚ฌ์šฉํ•˜์—ฌ ์ˆ˜์ง‘๋˜์—ˆ์Šต๋‹ˆ๋‹ค.
**How can the owner/curator/manager of the dataset be contacted?** Contact MLCommons via the website.
**๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์˜ ์†Œ์œ ์ž/ํ๋ ˆ์ดํ„ฐ/๊ด€๋ฆฌ์ž๋Š” ์–ด๋–ป๊ฒŒ ์—ฐ๋ฝํ•  ์ˆ˜ ์žˆ๋‚˜์š”?** ์›น์‚ฌ์ดํŠธ๋ฅผ ํ†ตํ•ด MLCommons์— ๋ฌธ์˜ํ•˜์„ธ์š”.
**๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์˜ ์†Œ์œ ์ž/ํ๋ ˆ์ดํ„ฐ/๊ด€๋ฆฌ์ž๋Š” ์–ด๋–ป๊ฒŒ ์—ฐ๋ฝํ•  ์ˆ˜ ์žˆ๋‚˜์š”?** ์›น์‚ฌ์ดํŠธ๋ฅผ ํ†ตํ•ด MLCommons์— ๋ฌธ์˜ํ•˜์„ธ์š”.
German Benchmarks from (Pluster, 2023), which translated their English counterparts using GPT 3.5 API
GPT 3.5 API๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์˜์–ด ๋ฒค์น˜๋งˆํฌ๋ฅผ ๋ฒˆ์—ญํ•œ (Pluster, 2023)์˜ ๋…์ผ์–ด ๋ฒค์น˜๋งˆํฌ
GPT 3.5 API๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์˜์–ด ๋ฒค์น˜๋งˆํฌ๋ฅผ ๋ฒˆ์—ญํ•œ (Pluster, 2023)์˜ ๋…์ผ์–ด ๋ฒค์น˜๋งˆํฌ