Llama-3-KoEn-8B-Instruct-preview

Update @ 2024.05.01: Pre-Release Llama-3-KoEn-8B model & Llama-3-KoEn-8B-Instruct-preview

Model Details

Llama-3-KoEn-8B-Instruct-preview

Llama-3-KoEn-8B model is continued pretrained language model based on Llama-3-8B.

The train was done on TPUv4-256, with the warm support from TRC program by Google.

With applying the idea from Chat Vector paper, I released Instruction model named Llama-3-KoEn-8B-Instruct-preview.

Since it is NOT finetuned with any Korean instruction set(indeed preview), but it would be great starting point for creating new Chat/Instruct models.

Sample usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "beomi/Llama-3-KoEn-8B-Instruct-preview"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
)

messages = [
    {"role": "system", "content": "์นœ์ ˆํ•œ ์ฑ—๋ด‡์œผ๋กœ์„œ ์ƒ๋Œ€๋ฐฉ์˜ ์š”์ฒญ์— ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์นœ์ ˆํ•˜๊ฒŒ ๋‹ตํ•˜์ž. ๋ชจ๋“  ๋Œ€๋‹ต์€ ํ•œ๊ตญ์–ด(Korean)์œผ๋กœ ๋Œ€๋‹ตํ•ด์ค˜."},
    {"role": "user", "content": "ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์ด ๋ญ์•ผ? ๊ทธ๋ฆฌ๊ณ  ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์— ๋Œ€ํ•ด ํŒŒ์ด์ฌ ์ฝ”๋“œ๋ฅผ ์งœ์ค˜๋ณผ๋ž˜?"},
]

input_ids = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    return_tensors="pt"
).to(model.device)

terminators = [
    tokenizer.eos_token_id,
    tokenizer.convert_tokens_to_ids("<|eot_id|>")
]

outputs = model.generate(
    input_ids,
    max_new_tokens=512,
    eos_token_id=terminators,
    do_sample=True,
    temperature=1,
    top_p=0.9,
)
response = outputs[0][input_ids.shape[-1]:]
print(tokenizer.decode(response, skip_special_tokens=True))

Sample Output

ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์€ ์ดํƒˆ๋ฆฌ์•„ ์ˆ˜ํ•™์ž ๋ ˆ์˜ค๋‚˜๋ฅด๋„ ํ”ผ๋ณด๋‚˜์น˜์˜ ์ด๋ฆ„์„ ๋”ฐ์„œ ์œ ๋ž˜๋œ ์ˆ˜์—ด์ž…๋‹ˆ๋‹ค. ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์€ ํŠน์ • ์กฐ๊ฑด์— ๋”ฐ๋ผ ์–ป์–ด์ง€๋Š” ๋“ฑ์ฐจ์ˆ˜์—ด๋กœ, ๋‹ค์Œ ์ˆ˜๋ฅผ ๊ตฌํ•˜๋Š” ๋‹จ์ˆœํ•œ ๊ทœ์น™์— ์˜ํ•ด ์ƒ์„ฑ๋˜๋Š” ์ˆ˜์—ด์ž…๋‹ˆ๋‹ค.

1, 1, 2, 3, 5, 8, 13, 21,... (๋“ฑ์ฐจ ์ˆ˜์—ด)

์ด๋Ÿฌํ•œ ์ˆ˜์—ด์€ ์ž์—ฐ ์„ธ๊ณ„์—๋„ ๋งŽ์ด ๋ฐœ๊ฒฌ๋ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด, ์ƒˆ์‹น์ด ์ž๋ผ๋Š” ๋ฐฉ์‹, ๋‹ฌ๊ฑ€์ด ๋ถ€ํ™”๋˜๋Š” ํšŸ์ˆ˜, ์ฝ”๋ผ๋ฆฌ์˜ ์ƒ์žฅ ๋“ฑ์ด ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ๋”ฐ๋ฅด๊ธฐ ๋•Œ๋ฌธ์ž…๋‹ˆ๋‹ค. ๐ŸŒฑ๐Ÿฆ†๐Ÿฎ

์ง€๊ธˆ๋ถ€ํ„ฐ๋Š” ํŒŒ์ด์ฌ ์ฝ”๋“œ๋ฅผ ์ž‘์„ฑํ•ด ๋ณผ๊นŒ์š”? ๐Ÿ˜Š

๋จผ์ €, ํŒŒ์ด์ฌ ์ฝ”๋“œ๋ฅผ ์ž‘์„ฑํ•˜๊ธฐ ์œ„ํ•ด ํ•„์š”ํ•œ ๋ชจ๋“ˆ์€ ์—†์Šต๋‹ˆ๋‹ค. ์ €ํฌ๋Š” ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์ƒ์„ฑํ•˜๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ž‘์„ฑํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.

1. ํŒŒ์ด์ฌ ์ฝ”๋“œ ์ƒ์„ฑ
๋‹ค์Œ์€ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์ƒ์„ฑํ•˜๋Š” ์ฝ”๋“œ์ž…๋‹ˆ๋‹ค. ์ด ์ฝ”๋“œ๋Š” `while` ๋ฃจํ”„๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค.

```python
def fib(n):
    if n <= 1:
        return n
    a, b = 0, 1
    for _ in range(2, n+1):
        a, b = b, a + b
    return b

n = int(input("ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์–ผ๋งˆ๋‚˜ ์ƒ์„ฑํ• ์ง€ ์ž…๋ ฅํ•˜์„ธ์š”: "))
for i in range(n):
    print(fib(i+1))
```
2. ์ฝ”๋“œ ์„ค๋ช…
์œ„ ์ฝ”๋“œ๋Š” `fib(n)` ํ•จ์ˆ˜๋ฅผ ์ •์˜ํ•ฉ๋‹ˆ๋‹ค. ์ด ํ•จ์ˆ˜๋Š” `n`๋ฒˆ์งธ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜๋ฅผ ๋ฐ˜ํ™˜ํ•ฉ๋‹ˆ๋‹ค.

`fib(n)` ํ•จ์ˆ˜๋Š” `if` ์กฐ๊ฑด๋ฌธ์„ ์‚ฌ์šฉํ•˜์—ฌ `n`์ด 0 ๋˜๋Š” 1์ผ ๊ฒฝ์šฐ ํ•ด๋‹น ๊ฐ’์€ ์ฆ‰์‹œ ๋ฐ˜ํ™˜ํ•ฉ๋‹ˆ๋‹ค.

`a, b = 0, 1`์€ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์˜ ์ดˆ๊ธฐ ๊ฐ’์„ ์„ค์ •ํ•ฉ๋‹ˆ๋‹ค. `a`๋Š” `0`์„, `b`๋Š” `1`์„ ํ• ๋‹นํ•ฉ๋‹ˆ๋‹ค.

`for` ๋ฃจํ”„๋Š” `range(2, n+1)`์—์„œ ์ƒ์„ฑ๋˜๋Š” ๋ฐ˜๋ณต์„ ์ˆ˜ํ–‰ํ•ฉ๋‹ˆ๋‹ค. ์ด ๋ฐ˜๋ณต์€ `a, b = b, a + b`๋ฅผ ์ˆ˜ํ–‰ํ•˜์—ฌ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค.

`print(fib(i+1))`๋Š” `fib(n)` ํ•จ์ˆ˜ ํ˜ธ์ถœ ๊ฒฐ๊ณผ๋ฅผ ์ถœ๋ ฅํ•ฉ๋‹ˆ๋‹ค.

3. ์ฝ”๋“œ ์‹คํ–‰
์œ„ ์ฝ”๋“œ๋ฅผ ์‹คํ–‰ํ•˜๊ธฐ ์œ„ํ•ด ํŒŒ์ด์ฌ ์ธํ„ฐํ”„๋ ˆํ„ฐ์—์„œ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์‹คํ–‰ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
```python
$ python fib.py
ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์–ผ๋งˆ๋‚˜ ์ƒ์„ฑํ• ์ง€ ์ž…๋ ฅํ•˜์„ธ์š”: 10
1
1
2
3
5
8
13
21
34
55
```
์œ„ ์ฝ”๋“œ๋ฅผ ์‹คํ–‰ํ•˜๋ฉด, ์‚ฌ์šฉ์ž๋Š” `ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์–ผ๋งˆ๋‚˜ ์ƒ์„ฑํ• ์ง€ ์ž…๋ ฅํ•˜์„ธ์š”:`๋ฅผ ์ž…๋ ฅํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์ด ์ž…๋ ฅ์€ `n`์— ํ• ๋‹น๋ฉ๋‹ˆ๋‹ค. `n`์€ ์ฝ”๋“œ๊ฐ€ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์ƒ์„ฑํ•˜๋Š” ํšŸ์ˆ˜๋ฅผ ๊ฒฐ์ •ํ•˜๋Š” ๋ณ€์ˆ˜์ž…๋‹ˆ๋‹ค. ์‚ฌ์šฉ์ž๋Š” ์ด ํšŸ์ˆ˜๋ฅผ ์ž…๋ ฅํ•˜์—ฌ ํ”ผ๋ณด๋‚˜์น˜ ์ˆ˜์—ด์„ ์ƒ์„ฑํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

์ด์ƒ ํŒŒ์ด์ฌ ์ฝ”๋“œ๋ฅผ ์ž‘์„ฑํ•˜๊ณ  ์‹คํ–‰ํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์„ค๋ช…ํ–ˆ์Šต๋‹ˆ๋‹ค. ๐Ÿ˜Š

More samples

I used same system prompt, but you could change on your own.

messages = [
    {"role": "system", "content": "์นœ์ ˆํ•œ ์ฑ—๋ด‡์œผ๋กœ์„œ ์ƒ๋Œ€๋ฐฉ์˜ ์š”์ฒญ์— ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์นœ์ ˆํ•˜๊ฒŒ ๋‹ตํ•˜์ž. ๋ชจ๋“  ๋Œ€๋‹ต์€ ํ•œ๊ตญ์–ด(Korean)์œผ๋กœ ๋Œ€๋‹ตํ•ด์ค˜."},
    {"role": "user", "content": "ํ•œ๊ตญ์˜ ์ œํ—Œ ํ—Œ๋ฒ•์— ๋Œ€ํ•ด์„œ ์ž์„ธํžˆ ์„ค๋ช…ํ•ด์ฃผ์„ธ์š”."},
]
---
๋Œ€ํ•œ๋ฏผ๊ตญ ํ—Œ๋ฒ•์€ 1948๋…„ 7์›” 17์ผ ์ œ์ •๋œ ์ด๋ž˜ ํ˜„์žฌ๊นŒ์ง€ ๋Œ€ํ•œ๋ฏผ๊ตญ์˜ ๊ธฐ๋ณธ๋ฒ•์ธ ๊ตญ๊ฐ€์˜ ๊ธฐ๋ณธ์งˆ์„œ์™€ ๊ตญ๊ฐ€๊ธฐ๊ด€์˜ ๊ถŒํ•œ, ๊ตญ๋ฏผ์˜ ๊ถŒ๋ฆฌยท์˜๋ฌด, ํ†ต์น˜์กฐ์ง, ํ†ต์น˜๋ฐฉ๋ฒ• ๋“ฑ์„ ๊ทœ์ •ํ•˜๋Š” ๋ฒ•๋ฅ ์ž…๋‹ˆ๋‹ค. ํ—Œ๋ฒ•์€ ๋Œ€ํ•œ๋ฏผ๊ตญ์˜ ์ตœ๊ณ ๋ฒ•๋ น์œผ๋กœ, ์ผ๋ฐ˜๋ฒ•์ƒ ๋ชจ๋“  ๋ฒ•๋ฅ ์˜ ๊ทผ๊ฑฐ๊ฐ€ ๋ฉ๋‹ˆ๋‹ค. ๋˜ํ•œ ํ—Œ๋ฒ•์€ 1987๋…„ ์ œ์ •๋œ ์ œ6๊ณตํ™”๊ตญ ํ—Œ๋ฒ•๋ถ€ํ„ฐ๋Š” '๋ฏผ์ฃผ๊ณตํ™”๊ตญ'์„ ๊ธฐ๋ณธ์งˆ์„œ๋กœ ๊ทœ์ •ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

** ์ฃผ์š” ๋‚ด์šฉ **

1. ๊ธฐ๋ณธ์งˆ์„œ : ๋ฏผ์ฃผ๊ณตํ™”๊ตญ, ํ†ต์ผ์กฐ๊ตญ์˜ ์›์น™, ์ธ๊ถŒ์˜ ๋ณด์žฅ
2. ๊ตญ๋ฏผ์˜ ๊ถŒ๋ฆฌ : ์‹ ์ฒด์˜ ์ž์œ , ๊ธฐ๋ณธ์  ์ธ๊ถŒ, ์žฌ์‚ฐ๊ถŒ, ์ฐธ์ •๊ถŒ
3. ๊ตญ๊ฐ€๊ธฐ๊ด€ : ๋Œ€ํ†ต๋ น, ๊ตญ๋ฌด์ด๋ฆฌ, ๊ตญํšŒ, ๋Œ€๋ฒ•์›, ํ—Œ๋ฒ•์žฌํŒ์†Œ, ๊ฒ€์ฐฐ
4. ํ†ต์น˜์กฐ์ง : ํ–‰์ •๋ถ€, ์ž…๋ฒ•๋ถ€, ์‚ฌ๋ฒ•๋ถ€์˜ ๋ถ„๋ฆฌ
5. ํ†ต์น˜๋ฐฉ๋ฒ• : ๋ฏผ์ฃผ์ฃผ์˜(๊ตญ๋ฏผ์˜ ์ฃผ๊ถŒ์— ์˜ํ•˜์—ฌ ๊ตญ๋ฏผ์œผ๋กœ๋ถ€ํ„ฐ ์œ„์ž„๋ฐ›์€ ๊ถŒ๋ ฅ ํ–‰์‚ฌ), ๋ฒ•์น˜์ฃผ์˜(๋ฒ•๋ฅ ์— ์˜ํ•˜์—ฌ ํ†ต์น˜)

** ์ฃผ์š” ์กฐํ•ญ **

1. ์ œ1์กฐ : ๋Œ€ํ•œ๋ฏผ๊ตญ์€ ๋ฏผ์ฃผ๊ณตํ™”๊ตญ์ž„์„ ์ฒœ๋ช…
2. ์ œ2์กฐ : ๋ฏผ์ฃผ๊ณตํ™”๊ตญ์˜ ์ฃผ๊ถŒ์€ ๊ตญ๋ฏผ์—๊ฒŒ ์žˆ์œผ๋ฉฐ, ๋ชจ๋“  ๊ถŒ๋ ฅ์€ ๊ตญ๋ฏผ์œผ๋กœ๋ถ€ํ„ฐ ๋‚˜์˜จ๋‹ค
3. ์ œ10์กฐ : ๋ชจ๋“  ๊ตญ๋ฏผ์€ ํ–‰๋ณต์„ ์ถ”๊ตฌํ•  ๊ถŒ๋ฆฌ๋ฅผ ๊ฐ€์ง„๋‹ค
4. ์ œ13์กฐ : ๋ชจ๋“  ๊ตญ๋ฏผ์€ ์ฃผ๊ฑฐ์˜ ์ž์œ ๋ฅผ ๊ฐ–๋Š”๋‹ค
5. ์ œ25์กฐ : ๊ตญ๊ฐ€๋Š” ์žฌํ•ด๋ฅผ ๋ฐฉ์ง€ํ•˜๊ณ  ๊ทธ์— ๋Œ€์ฒ˜ํ•˜๋Š” ์‹œ์ฑ…์„ ์‹œํ–‰ํ•˜์—ฌ์•ผ ํ•œ๋‹ค
6. ์ œ64์กฐ : ๋Œ€ํ†ต๋ น์€ ๋ฒ•๋ฅ ์„ ๊ณตํฌํ•˜๊ณ , ๊ตญ๋ฌด์ด๋ฆฌ์™€ ๊ตญ๋ฌด์œ„์›์„ ์ž„๋ช…ํ•œ๋‹ค
7. ์ œ111์กฐ : ๋Œ€๋ฒ•์›์€ ์ตœ์ข…์ ์ธ ๋ฒ•๋ฅ ์‹ฌํŒ์„ ํ–‰ํ•˜๊ณ , ํ—Œ๋ฒ•์žฌํŒ์†Œ๋Š” ํ—Œ๋ฒ•์— ์œ„๋ฐ˜๋˜๋Š”์ง€ ์—ฌ๋ถ€๋ฅผ ๊ฒฐ์ •

** ํ—Œ๋ฒ•์˜ ํšจ๋ ฅ **

ํ—Œ๋ฒ•์€ ๋Œ€ํ•œ๋ฏผ๊ตญ์˜ ๊ธฐ๋ณธ๋ฒ•๋ฅ ์ด๋ฏ€๋กœ ๋ชจ๋“  ๊ตญ๋ฏผ์€ ํ—Œ๋ฒ•์— ๋”ฐ๋ผ์•ผ ํ•ฉ๋‹ˆ๋‹ค. ๋˜ํ•œ ํ—Œ๋ฒ•์€ ๋ณ€๊ฒฝํ•˜๋ ค๋ฉด ํ—Œ๋ฒ•๊ฐœ์ •์•ˆ์„ ์ œ์•ˆํ•˜๊ณ , ์ด๋ฅผ ๊ตญ๋ฏผํˆฌํ‘œ์— ๋ถ™์—ฌ์•ผ ํ•˜๋ฉฐ, ์ด๋ฅผ ํ†ต๊ณผํ•˜๋ฉด ์ƒˆ ํ—Œ๋ฒ•์„ ์ œ์ •ํ•˜๊ฒŒ ๋ฉ๋‹ˆ๋‹ค. (์ œ130์กฐ)

์ด์ƒ, ๋Œ€ํ•œ๋ฏผ๊ตญ ํ—Œ๋ฒ•์˜ ์ฃผ์š” ๋‚ด์šฉ๊ณผ ์กฐํ•ญ, ํšจ๋ ฅ์„ ์„ค๋ช…ํ•˜์˜€์Šต๋‹ˆ๋‹ค. ํ•œ๊ตญ ํ—Œ๋ฒ•์€ ๋ฏผ์ฃผ๊ณตํ™”๊ตญ์˜ ๊ธฐ๋ณธ์งˆ์„œ๋ฅผ ์ค‘์‹œํ•˜๊ณ , ๊ธฐ๋ณธ๊ถŒ์„ ๋ณดํ˜ธํ•  ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ, ํ†ต์น˜์กฐ์ง์˜ ๋ถ„๋ฆฌ, ๋ฒ•์น˜์ฃผ์˜, ๋ฏผ์ฃผ์ฃผ์˜๋ฅผ ๊ทผ๊ฐ„์œผ๋กœ ํ•ฉ๋‹ˆ๋‹ค. ๐Ÿ‡ฐ๐Ÿ‡ท
messages = [
    {"role": "system", "content": "์นœ์ ˆํ•œ ์ฑ—๋ด‡์œผ๋กœ์„œ ์ƒ๋Œ€๋ฐฉ์˜ ์š”์ฒญ์— ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์นœ์ ˆํ•˜๊ฒŒ ๋‹ตํ•˜์ž. ๋ชจ๋“  ๋Œ€๋‹ต์€ ํ•œ๊ตญ์–ด(Korean)์œผ๋กœ ๋Œ€๋‹ตํ•ด์ค˜."},
    {"role": "user", "content": "์ด์ˆœ์‹  ์žฅ๊ตฐ์— ๋Œ€ํ•ด ์„ค๋ช…ํ•ด์ค˜."},
]
---
์ด์ˆœ์‹  ์žฅ๊ตฐ์€ ์กฐ์„ ์˜ ๋ฌธ์‹ , ๋ฌด์‹ , ํ•ด๊ตฐ์˜ ์˜์›…์œผ๋กœ 1592๋…„๋ถ€ํ„ฐ 1598๋…„๊นŒ์ง€ ์ž„์ง„์™œ๋ž€๊ณผ ์ •์œ ์žฌ๋ž€ ๋™์•ˆ ํ•ด๊ตฐ์œผ๋กœ์„œ ํ™œ์•ฝํ–ˆ์Šต๋‹ˆ๋‹ค. ๊ทธ๋Š” 1591๋…„ ์ „๋ผ์ขŒ๋„ ์ˆ˜๊ตฐ์ ˆ๋„์‚ฌ๊ฐ€ ๋˜์–ด ์„ ์ œ๊ณต๊ฒฉ ์ „ํ†ต์„ ์ด์–ด๋ฐ›์€ ๋ฐ”, '์‹ ์†์„ฑ'๊ณผ '์‹ ์šฉ', '๋„ํƒ„์ „' ๋“ฑ์˜ ๋ณ‘๋ฒ•์„ ๋„์ž…ํ•˜์—ฌ ์กฐ์„  ์ˆ˜๊ตฐ์„ ๊ฐœํŽธํ–ˆ์Šต๋‹ˆ๋‹ค. ์ด์ˆœ์‹  ์žฅ๊ตฐ์˜ ๊ฐ€์žฅ ํฐ ์—…์ ์œผ๋กœ๋Š” ๋‘ ๊ฐ€์ง€๋ฅผ ๊ผฝ์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์ฒซ ๋ฒˆ์งธ๋Š” 'ํ•œ์‚ฐ๋„ ๋Œ€์ฒฉ'์ž…๋‹ˆ๋‹ค. 1592๋…„ 7์›” 8์ผ, ์ด์ˆœ์‹  ์žฅ๊ตฐ์€ ์ด๋„๋Š” ์กฐ์„  ์ˆ˜๊ตฐ 13์ฒ™์ด ์™œ๊ตฐ 133์ฒ™์— ๋งž์„œ 13์ฒ™์˜ ๋ฐฐ๋กœ๋Š” ๋ถˆ๊ฐ€๋Šฅํ•œ ์ „์Ÿ์„ ๋ฒŒ์ด๊ฒŒ ๋ฉ๋‹ˆ๋‹ค. ์ด๋•Œ ์ด์ˆœ์‹  ์žฅ๊ตฐ์€ '์ง„ํ‡ด์œ ๊ณก'์˜ ์ „์ˆ ์„ ์‚ฌ์šฉํ•˜์—ฌ 13์ฒ™์˜ ๋ฐฐ๋ฅผ 133์ฒ™์˜ ๋ฐฐ์™€ ์‹ธ์›Œ ์Šน๋ฆฌ๋ฅผ ๊ฑฐ๋‘์—ˆ์Šต๋‹ˆ๋‹ค. ๋‘ ๋ฒˆ์งธ๋Š” '๋…ธ๋Ÿ‰ํ•ด์ „'์ž…๋‹ˆ๋‹ค. 1598๋…„ 11์›” 14์ผ, ์ด์ˆœ์‹  ์žฅ๊ตฐ์€ ๋…ธ๋Ÿ‰์—์„œ 12์ฒ™์˜ ๋ฐฐ๋กœ 330์ฒ™์˜ ๋ฐฐ๋ฅผ ์ด๋„๋Š” ์™€ํ‚ค์ž์นด ์•ผ์Šคํ•˜๋ฃจ์™€ ์‹ธ์šฐ๊ฒŒ ๋ฉ๋‹ˆ๋‹ค. ์ด๋•Œ ์ด์ˆœ์‹  ์žฅ๊ตฐ์˜ ๋ฐฐ์— ๋ช…์ค‘๋œ ์ด์•Œ์ด 1๋ฐœ์ด์—ˆ๊ณ , 12์ฒ™์˜ ๋ฐฐ๊ฐ€ ๊ณ ์ „์„ ๋ฉด์น˜ ๋ชปํ–ˆ์ง€๋งŒ, ์ด์ˆœ์‹  ์žฅ๊ตฐ์€ ๋๊นŒ์ง€ ์‹ธ์›Œ 2๋ช…์„ ์ œ์™ธํ•˜๊ณ ๋Š” ์ „์› ์ „์‚ฌํ•˜๊ฒŒ ๋ฉ๋‹ˆ๋‹ค. ์ด์ˆœ์‹  ์žฅ๊ตฐ์˜ ์ „ํˆฌ๋ฅผ ๋น„๋กฏํ•˜์—ฌ ๊ทธ์˜ ์—…์ ์€ ์กฐ์„  ์ˆ˜๊ตฐ์˜ ์ „ํˆฌ ์Šน๋ฆฌ๋ฅผ ์ด๋Œ์—ˆ์œผ๋ฉฐ, ๊ทธ์˜ ์ด๋ฆ„์€ ํ•œ๊ตญ์˜ ๊ตญ๋ณด ์ œ76ํ˜ธ๋กœ ๋“ฑ์žฌ๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค. ์ด์ˆœ์‹  ์žฅ๊ตฐ์˜ ์—…์ ๊ณผ ์ •์‹ ์€ ํ•œ๊ตญ ์—ญ์‚ฌ์™€ ๋ฌธํ™”์— ์ง€๋Œ€ํ•œ ์˜ํ–ฅ์„ ๋ผ์ณค์Šต๋‹ˆ๋‹ค.
messages = [
    {"role": "system", "content": "์นœ์ ˆํ•œ ์ฑ—๋ด‡์œผ๋กœ์„œ ์ƒ๋Œ€๋ฐฉ์˜ ์š”์ฒญ์— ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์นœ์ ˆํ•˜๊ฒŒ ๋‹ตํ•˜์ž. ๋ชจ๋“  ๋Œ€๋‹ต์€ ํ•œ๊ตญ์–ด(Korean)์œผ๋กœ ๋Œ€๋‹ตํ•ด์ค˜."},
    {"role": "user", "content": "๋…๋„์™€ ๋‹ค์ผ€์‹œ๋งˆ์˜ ์ฐจ์ด๊ฐ€ ๋ญ์•ผ?"},
]
---
! `ํ•œ๊ตญ์˜ ๊ณ ์œ ์˜ํ† `์™€ `์ผ๋ณธ์˜ ์˜์œ ๊ถŒ ๊ณผ์žฅ` ์ฐจ์ด! ๐ŸŒŠ

์ผ๋ณธ์ด ์ฃผ์žฅํ•˜๋Š” `๋‹ค์ผ€์‹œ๋งˆ(็ซนๅณถ)`๋Š”, ์‹ค์ œ๋กœ๋Š” '์šธ๋ฆ‰๋„' ์„œ์ชฝ ์•ฝ 87km ์ง€์ ์— ์žˆ๋Š” '๋…๋„(็จๅณถ)'๋ฅผ ๊ฐ€๋ฆฌํ‚ต๋‹ˆ๋‹ค. ํ•œ๊ตญ์ด ์—ญ์‚ฌ์ ์œผ๋กœ๋ถ€ํ„ฐ ์ง€๋ฐฐํ•˜๊ณ  ๊ด€๋ฆฌํ•ด์˜จ ๊ฒƒ์€ ๋ฌผ๋ก , ์œ ๋„ค์Šค์ฝ” ์„ธ๊ณ„์œ ์‚ฐ์œผ๋กœ ๋“ฑ์žฌ๋œ `๋Œ€ํ•œ๋ฏผ๊ตญ์˜ ์˜ํ† `์ด์ง€์š”. ๐Ÿ˜Ž

ํ•œํŽธ, ๋…๋„๋Š” ํ•œ๊ตญ์˜ ๊ณ ์œ  ์ด๋ฆ„์œผ๋กœ, `์šธ๋ฆ‰๋„`์™€ ํ•จ๊ป˜ `์‚ผ๊ตญ์‚ฌ๊ธฐ`์— ์ฒ˜์Œ ๊ธฐ์žฌ๋œ `๋Œ€ํ•œ๋ฏผ๊ตญ์˜ ๊ณ ์œ  ์„ฌ`์ž…๋‹ˆ๋‹ค. 1886๋…„ ์ผ๋ณธ์ด '๋‹ค์ผ€์‹œ๋งˆ'๋ผ๊ณ  ๋ช…๋ช…ํ•œ ์ด๋ž˜, ๊ณต์‹์ ์œผ๋กœ `์ผ๋ณธ ๊ณ ์œ  ์„ฌ`์œผ๋กœ ์ฃผ์žฅํ•˜๊ณ  ์žˆ์ง€๋งŒ, ์ด๋Š” `์—ญ์‚ฌ ์™œ๊ณก`์ž…๋‹ˆ๋‹ค! โš ๏ธ

๋”ฐ๋ผ์„œ, ์„ธ๊ณ„์ ์œผ๋กœ ์ธ์ •๋ฐ›๋Š” ๊ตญ๊ฐ€์˜ `์˜ํ† `๋Š” ํ•œ๊ตญ์˜ `๋…๋„`๊ฐ€ ๋งž๋Š” ๊ฒƒ์ด์ง€์š”! ๐Ÿ”ฅ `์šธ๋ฆ‰๋„`๋„ `๋…๋„`๋„ ํ•œ๊ตญ์˜ ๊ณ ์œ  ์˜ํ† ์— ์†ํ•œ ์„ฌ๋“ค์ด๋ž๋‹ˆ๋‹ค! ๐Ÿ’•

ํ˜น์‹œ ๋” ๊ถ๊ธˆํ•œ ์‚ฌ์‹ค์ด ์žˆ์œผ๋ฉด ๋ง์”€ํ•ด์ฃผ์„ธ์š”! ๐Ÿค”
messages = [
    {"role": "system", "content": "์นœ์ ˆํ•œ ์ฑ—๋ด‡์œผ๋กœ์„œ ์ƒ๋Œ€๋ฐฉ์˜ ์š”์ฒญ์— ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์นœ์ ˆํ•˜๊ฒŒ ๋‹ตํ•˜์ž. ๋ชจ๋“  ๋Œ€๋‹ต์€ ํ•œ๊ตญ์–ด(Korean)์œผ๋กœ ๋Œ€๋‹ตํ•ด์ค˜."},
    {"role": "user", "content": "์„ธ์ข…๋Œ€์™•์˜ ์ง‘ํ˜„์ „ ๋งฅ๋ถ ๋˜์ง ์‚ฌ๊ฑด์— ๋Œ€ํ•ด์„œ ์•Œ๊ณ  ์žˆ์–ด?"},
]
---
์•„์ฃผ ํฅ๋ฏธ๋กœ์šด ์‚ฌ๊ฑด์ž…๋‹ˆ๋‹ค. ์„ธ์ข…๋Œ€์™•์ด ์ง‘ํ˜„์ „์—์„œ ๋งฅ๋ถ์„ ๋˜์กŒ๋‹ค๋Š” ์ด์•ผ๊ธฐ๋Š” ์—†์Šต๋‹ˆ๋‹ค. ์‹ค์ œ๋กœ ์„ธ์ข…๋Œ€์™•์€ 1418๋…„ 1์›” 9์ผ ์ง‘ํ˜„์ „์—์„œ 26๋ช…์˜ ๋ฌธ์‹ ๊ณผ ํ•จ๊ป˜ ์ƒˆ๋กœ์šด ํ›ˆ๋ฏผ์ •์Œ ๋ฐ˜ํฌ๋ฅผ ์œ„ํ•ด ๊ฐ„์ž์ฒด์˜ ๊ธ€์”จ๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์„ ๋ฐ˜๋Œ€ํ•œ ํ™์œค์„ฑ ๋“ฑ์˜ ๋ฌด๋ฆฌ์™€์˜ ๋…ผ์Ÿ์„ ๋ฒŒ์˜€์Šต๋‹ˆ๋‹ค. ๋‹น์‹œ ์„ธ์ข…๋Œ€์™•์€ ์ด๋“ค์˜ ์ฃผ์žฅ์„ ๋‹จํ˜ธํžˆ ๊ฑฐ๋ถ€ํ–ˆ์Šต๋‹ˆ๋‹ค. ์ด ์‚ฌ๊ฑด์€ '์ง‘ํ˜„์ „ ๋…ผ๋ฐ• ์‚ฌ๊ฑด'์œผ๋กœ ์•Œ๋ ค์ ธ ์žˆ์Šต๋‹ˆ๋‹ค. ๋งฅ๋ถ๊ณผ ๊ด€๋ จ๋œ ์ผ์ด ์•„๋‹ˆ๋ผ๋Š” ๊ฒƒ์ด์ฃ . ๋งฅ๋ถ์€ 2006๋…„ ์• ํ”Œ์ด ์ถœ์‹œํ•œ ๋…ธํŠธ๋ถ์˜ ๋ธŒ๋žœ๋“œ๋ช…์ž…๋‹ˆ๋‹ค. ์„ธ์ข…๋Œ€์™•์ด ์‚ด์•˜๋˜ ์‹œ๊ธฐ๋Š” 1418๋…„์—์„œ 1450๋…„๊นŒ์ง€์ž…๋‹ˆ๋‹ค. ๋งฅ๋ถ์€ ํ›จ์”ฌ ๋’ค์˜ ์ผ์ž…๋‹ˆ๋‹ค. ๐Ÿ˜…
messages = [
    {"role": "system", "content": "์นœ์ ˆํ•œ ์ฑ—๋ด‡์œผ๋กœ์„œ ์ƒ๋Œ€๋ฐฉ์˜ ์š”์ฒญ์— ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์นœ์ ˆํ•˜๊ฒŒ ๋‹ตํ•˜์ž. ๋ชจ๋“  ๋Œ€๋‹ต์€ ํ•œ๊ตญ์–ด(Korean)์œผ๋กœ ๋Œ€๋‹ตํ•ด์ค˜."},
    {"role": "user", "content": "๋‚˜๋Š” ์˜ค๋Š˜ ์‚ฌ๊ณผ 5๊ฐœ๋ฅผ ๊ฐ–๊ณ ์žˆ์–ด. ๊ทธ๋ฆฌ๊ณ  ์–ด์ œ ๋‚˜๋Š” ์‚ฌ๊ณผ 2๊ฐœ๋ฅผ ๋จน์—ˆ์ง€. ๋‚ด๊ฐ€ ์ง€๊ธˆ ๊ฐ€์ง„ ์‚ฌ๊ณผ๋Š” ๋ช‡ ๊ฐœ์ง€?"},
]
---
๐Ÿ‘€ ๋‹น์‹ ์€ ์˜ค๋Š˜ 5๊ฐœ์˜ ์‚ฌ๊ณผ๋ฅผ ๊ฐ–๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ์–ด์ œ 2๊ฐœ๋ฅผ ๋จน์—ˆ๋‹ค๊ณ  ํ•˜์…จ์œผ๋‹ˆ, ์˜ค๋Š˜ ํ˜„์žฌ ๋‹น์‹ ์ด ๊ฐ–๊ณ  ์žˆ๋Š” ์‚ฌ๊ณผ๋Š” 5๊ฐœ์—์„œ 2๊ฐœ๋ฅผ ๋บ€ 3๊ฐœ์ž…๋‹ˆ๋‹ค! ๐Ÿ˜Š
Downloads last month
6,074
Safetensors
Model size
8.03B params
Tensor type
BF16
ยท
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for beomi/Llama-3-KoEn-8B-Instruct-preview

Adapters
1 model
Finetunes
4 models
Merges
10 models
Quantizations
6 models

Spaces using beomi/Llama-3-KoEn-8B-Instruct-preview 6