Edit model card

image/jpeg

This model was developed using the Mistral-7b base model. The name โ€œMistralโ€ is derived from an anti-aircraft weapon, which inspired the creation of Koreaโ€™s anti-aircraft weapon, Singung. Just as the anti-aircraft weapon was named after its inspiration, this model was named โ€œSingungโ€ because it uses the Mistral model.

The Lora tuning technique was used and is included in the weights.

SYSTEM_PROMPT = "### System:\n ๋‹น์‹ ์€ ์ฐจ๊ทผ์ฐจ๊ทผ ์ƒ๊ฐํ•˜๊ณ , ๋…ผ๋ฆฌ์ ์ธ ์‚ฌ๊ณ ๋ฅผ ํ†ตํ•ด ์ฃผ์–ด์ง„ ๋ฌธ์ œ๋ฅผ ํ‘ธ๋Š” ์ธ๊ณต์ง€๋Šฅ ์ž…๋‹ˆ๋‹ค."
USER_PROMPT = "\n ### User: \n ๋ฏผ์˜์ด๋Š” ์–ด์ œ ์‚ฌ๊ณผ๋ฅผ 3๊ฐœ ๋“ค๊ณ  ์žˆ์—ˆ์Šต๋‹ˆ๋‹ค. ์˜ค๋Š˜ ๊ณ ์–‘์ด๊ฐ€ ์‚ฌ๊ณผ๋ฅผ ๋‹ฌ๋ผ๊ณ  ํ–ˆ๋Š”๋ฐ ์ฃผ์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค. ๋ฏผ์˜์ด๋Š” ๋‚˜์œ ์‚ฌ๋žŒ ์ž…๋‹ˆ๊นŒ?"
ASSISTANT_PROMPT = "\n ### Assistant: \n"
template = f"{SYSTEM_PROMPT}{USER_PROMPT}{ASSISTANT_PROMPT}"

inputs = tokenizer(template, retuen_tensors="pt").to(0) # ๋ชจ๋ธ๊ณผ ๋™์ผํ•œ GPU์— ์‚ฌ์šฉํ•ด์•ผํ•จ. ์—†์œผ๋ฉด to(0) ์ง€์šฐ์„ธ์š”.
response = model.generate(**inputs, max_new_tokens=2048)
print(tokenizer.batch_decode(response)[0])

### response:

<s> ### System:
๋‹น์‹ ์€ ์ฐจ๊ทผ์ฐจ๊ทผ ์ƒ๊ฐํ•˜๊ณ , ๋…ผ๋ฆฌ์ ์ธ ์‚ฌ๊ณ ๋ฅผ ํ†ตํ•ด ์ฃผ์–ด์ง„ ๋ฌธ์ œ๋ฅผ ํ‘ธ๋Š” ์ธ๊ณต์ง€๋Šฅ ์ž…๋‹ˆ๋‹ค.
 ### User: 
 ๋ฏผ์˜์ด๋Š” ์–ด์ œ ์‚ฌ๊ณผ๋ฅผ 3๊ฐœ ๋“ค๊ณ  ์žˆ์—ˆ์Šต๋‹ˆ๋‹ค. ์˜ค๋Š˜ ๊ณ ์–‘์ด๊ฐ€ ์‚ฌ๊ณผ๋ฅผ ๋‹ฌ๋ผ๊ณ  ํ–ˆ๋Š”๋ฐ ์ฃผ์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค. ๋ฏผ์˜์ด๋Š” ๋‚˜์œ ์‚ฌ๋žŒ ์ž…๋‹ˆ๊นŒ?
 ### Assistant: 
๋ฏผ์˜์ด๊ฐ€ ๊ณ ์–‘์ด๊ฐ€ ์‚ฌ๊ณผ๋ฅผ ์ข‹์•„ํ•˜๋Š” ๊ฒƒ์„ ์•Œ๊ณ  ์žˆ์Œ์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ  ์‚ฌ๊ณผ๋ฅผ ์ฃผ์ง€ ์•Š์•˜๊ธฐ ๋•Œ๋ฌธ์— ๊ณ ์–‘์ด๋ฅผ ๋ถˆ์พŒํ•˜๊ฒŒ ํ•˜๊ณ  ๋ถˆ์Œํ•˜๊ฒŒ ๋งŒ๋“ค์—ˆ๊ธฐ ๋•Œ๋ฌธ์— ๋ฏผ์˜์ด๋Š” ์•ฝ๊ฐ„ ๋‚˜์œ ์‚ฌ๋žŒ์œผ๋กœ ๊ฐ„์ฃผ๋  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
๊ทธ๋Ÿฌ๋‚˜ ์ด๊ฒƒ์ด ๋‹จ์ง€ ์‚ฌ๊ณผ์— ๋Œ€ํ•œ ํ–‰๋™์—๋งŒ ๊ตญํ•œ๋œ ๊ฒƒ์€ ์•„๋‹ˆ๋ผ๋Š” ์ ์„ ๊ณ ๋ คํ•˜๋Š” ๊ฒƒ์ด ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค.
์‚ฌ๋žŒ๋งˆ๋‹ค ๋‹ค๋ฅธ ๊ฐ€์น˜๊ด€๊ณผ ์‹ ๋…์„ ๊ฐ€์ง€๊ณ  ์žˆ์œผ๋ฉฐ, ์–ด๋–ค ์‚ฌ๋žŒ๋“ค์€ ๋ฐ˜๋ ค๋™๋ฌผ์ด ๋จน์„ ๊ฒƒ์„ ์ฃผ์ง€ ์•Š๋Š” ๊ฒƒ์„ ์ •๋‹นํ™”ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
๋”ฐ๋ผ์„œ ๋ฏผ์˜์ด๊ฐ€ ๋‚˜์œ ์‚ฌ๋žŒ์ด๋ผ๊ณ  ๋‹จ์ •์ ์œผ๋กœ ๋งํ•˜๊ธฐ ์ „์— ๊ทธ๋…€์˜ ์˜๋„์™€ ์‹ ๋…์„ ๊ณ ๋ คํ•˜๋Š” ๊ฒƒ์ด ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค.</s>
Downloads last month
1,386
Safetensors
Model size
7.24B params
Tensor type
F32
ยท
Inference API
Model is too large to load in Inference API (serverless). To try the model, launch it on Inference Endpoints (dedicated) instead.

Dataset used to train StatPan/singung-sft-v0.1