Transformers
GGUF
Korean
mistral
trl
unsloth
Inference Endpoints
conversational
Edit model card

  • Base Model : 'HuggingFaceH4/zephyr-7b-beta'

  • Purpose : '์–ผ๋งˆ๋“ ์ง€ ๋ฌผ์–ด๋ณด์„ธ์š”~! ํ’‹ํ’‹!'์ด ๋ง ๋์— ๋ถ™์œผ๋ฉฐ 'ํ•ด์š”'์ฒด๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ์ฑ—๋ด‡์„ ๊ตฌํ˜„ํ•˜๋ ค๊ณ  ํ•œ๋‹ค. ํ”„๋กœ์ ํŠธ ๋ชฉ์ ์ƒ RAG๋ฅผ ํ†ตํ•ด ํ’‹์‚ด ๋„๋ฉ”์ธ์— ๋Œ€ํ•œ ์ •๋ณด๋ฅผ ์ œ๊ณตํ•˜๋Š” 'ํ’‹ํ’‹์ด' ์ปจ์…‰์ด๊ธฐ์— ๋งํˆฌ ์„ค์ •์ด ์ด์™€ ๊ฐ™๋‹ค.

  • Method : Unsloth ํŒจํ‚ค์ง€๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ gpu ์ž์›์ด ์ดˆ๊ณผ๋˜์ง€ ์•Š๋„๋ก ํ•˜์˜€์œผ๋ฉฐ SFTrainer๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธํ›ˆ๋ จ์„ ์ง„ํ–‰ํ•จ.

  • Environ : Colab L4 GPU๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ง„ํ–‰ํ•˜์˜€์Šต๋‹ˆ๋‹ค.

  • About gguf : ํ’‹ํ’‹์ด ์ตœ์ข… ๋ฒ„์ „์„ ๋งŒ๋“ค๊ธฐ ์œ„ํ•ด ๊ฐ ๋‹จ๊ณ„๋ณ„ ๋ชจ๋ธ์„ ์ €์žฅํ•˜์˜€์Šต๋‹ˆ๋‹ค. ๋ชจ๋ธ ์–‘์žํ™”(Quantize) ์ข…๋ฅ˜๋กœ F16, Q8_0, Q5_K_M 3๊ฐ€์ง€๋กœ ์ €์žฅํ•˜์˜€์Šต๋‹ˆ๋‹ค.

Downloads last month
1,476
GGUF
Model size
7.24B params
Architecture
llama

5-bit

8-bit

16-bit

Inference API
Unable to determine this modelโ€™s pipeline type. Check the docs .

Datasets used to train Dongwookss/futfut_by_zephyr7b_gguf