TenOS-Ko-27B-v2

FINAL-Bench/Darwin-27B-Opus 기반 한국어 특화 SFT 모델입니다.

Method

• Base Model: Darwin-27B-Opus (Qwen3.5-27B family)
• Korean SFT: 한국어 문화, 역사, 법률, 경제, 사회 등 한국 특화 지식 중심 Supervised Fine-Tuning
• Thinking Mode: <think> 태그를 통한 Chain-of-Thought 추론 지원

Model Specifications

Property	Value
Architecture	Qwen3.5 Hybrid (64 layers)
Parameters	~27B
Context Length	262,144 tokens
Precision	BF16
License	Apache 2.0

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model = AutoModelForCausalLM.from_pretrained(
    "honey90/TenOS-Ko-27B-v2",
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("honey90/TenOS-Ko-27B-v2")

messages = [{"role": "user", "content": "한국의 전통 명절에 대해 설명해주세요."}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=1024, do_sample=False)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Acknowledgements

• FINAL-Bench — Darwin-27B-Opus base model
• Qwen Team — Qwen3.5 architecture