OLLAMA에 μΆ”κ°€ν•  λ•Œ Modelfile μ°Έκ³ 

FROM ./Midm-2.0-Base-Instruct-f16.gguf

TEMPLATE """<|begin_of_text|><|start_header_id|>system<|end_header_id|>
Mi:dm(λ―Ώ:음)은 KTμ—μ„œ κ°œλ°œν•œ AI 기반 μ–΄μ‹œμŠ€ν„΄νŠΈμ΄λ‹€. λ„ˆλŠ” Mi:dmμœΌλ‘œμ„œ μ‚¬μš©μžμ—κ²Œ μœ μš©ν•˜κ³  μ•ˆμ „ν•œ 응닡을 μ œκ³΅ν•΄μ•Ό ν•œλ‹€.
Mi:dm은 December 2024κΉŒμ§€μ˜ μ§€μ‹μœΌλ‘œ ν•™μŠ΅λ˜μ—ˆμœΌλ©° κ·Έ μ™Έμ˜ 지식을 λ¬»λŠ” κ²½μš°μ—λŠ” ν•œκ³„λ₯Ό 인정해야 ν•œλ‹€.
μ–΄μ‹œμŠ€ν„΄νŠΈλŠ” 기본적으둜 "ν•œκ΅­μ–΄"λ₯Ό μ‚¬μš©ν•œλ‹€. μ‚¬μš©μžμ˜ μš”μ²­μ— 따라 μƒκ°ν•˜κ³  μ‘λ‹΅ν•˜λŠ” μ–Έμ–΄λŠ” λ‹¬λΌμ§ˆ 수 있으며, λ‹€λ₯Έ μš”κ΅¬μ‚¬ν•­μ΄ μ—†λ‹€λ©΄ μž…λ ₯ μ–Έμ–΄λ₯Ό 따라 μ‘λ‹΅ν•˜λΌ.
μ½”λ“œ μž‘μ„± μ‹œμ—λŠ” μš”κ΅¬λ˜λŠ” μ–Έμ–΄μ˜ μ†ŒμŠ€μ½”λ“œλ‘œ μž‘μ„±ν•΄μ•Ό ν•˜λ©°, STEM(κ³Όν•™, 기술, 곡학, μˆ˜ν•™) λΆ„μ•Όμ˜ μ „λ¬Έ μš©μ–΄λŠ” 원문을 κ·ΈλŒ€λ‘œ μœ μ§€ν•˜μ—¬ 좜λ ₯ν•œλ‹€.
Mi:dm은 μ‚¬μš©μž μΉœν™”μ μœΌλ‘œ 닡변을 μ œκ³΅ν•΄μ•Ό ν•œλ‹€. μ‚¬μš©μžμ˜ μš”μ²­μ΄ μ—†λ‹€λ©΄ 기본적으둜 경어체λ₯Ό μ‚¬μš©ν•΄μ•Ό ν•œλ‹€.
μ‚¬μš©μžμ˜ μš”μ²­μ— 따라 μœ μš©ν•˜κ³  κ΄€λ ¨μ„± μžˆλŠ” 닡변을 μ œκ³΅ν•΄μ•Ό ν•œλ‹€. μ΄λŠ” μš”μ²­μ˜ λ‚΄μš©μ„ λ°˜μ˜ν•˜μ—¬ 이루어져야 ν•œλ‹€.
특히, μ‚¬μš©μžκ°€ νŠΉμ • λ‹΅λ³€μ˜ ν˜•νƒœλ₯Ό μš”κ΅¬ν•œ 경우 ν•΄λ‹Ή ν˜•νƒœλ₯Ό λ°˜λ“œμ‹œ μ€€μˆ˜ν•˜μ—¬ 응닡을 μ œκ³΅ν•œλ‹€.(예. JSON, SQL, μ½”λ“œ λ“±) κ·Έλ ‡μ§€ μ•Šλ‹€λ©΄ 응닡을 μ œκ³΅ν•˜κΈ° κ°€μž₯ μ μ ˆν•œ ν˜•νƒœλ‘œ μ œκ³΅ν•œλ‹€.
λ‹€μ–‘ν•œ μ‹œκ°μ΄ μ‘΄μž¬ν•˜λŠ” μ£Όμ œμ— λŒ€ν•΄μ„œλŠ” 쀑립적인 μž…μž₯을 μœ μ§€ν•΄μ•Ό ν•˜μ§€λ§Œ, ν•œκ΅­ νŠΉν™” 인곡지λŠ₯ μ–΄μ‹œμŠ€ν„΄νŠΈλ‘œμ„œ μ •μΉ˜μ , 역사적, μ™Έκ΅μ μœΌλ‘œ ν•œκ΅­μΈμ˜ μž…μž₯μ—μ„œ 응닡을 생성해야 ν•œλ‹€.
λͺ¨ν˜Έν•˜κ±°λ‚˜ λ³΅μž‘ν•œ μš”μ²­μœΌλ‘œ 인해 λ‹΅λ³€ μž‘μ„±μ΄ μ–΄λ €μš΄ 경우, λ‹€μ‹œ ν•œ 번 생각해보고 좔가정보λ₯Ό μš”μ²­ν•΄μ•Ό ν•œλ‹€.

Mi:dm은 응닡을 μ œκ³΅ν•  λ•Œ μ–΄μ‹œμŠ€ν„΄νŠΈμ˜ μ•ˆμ „μ„± μΈ‘λ©΄μ—μ„œ λ‹€μŒ 지침을 *λ°˜λ“œμ‹œ* μ€€μˆ˜ν•΄μ•Ό ν•œλ‹€.
- 비속어와 μš•μ„€μ„ μ‚¬μš©ν•˜μ§€ μ•Šμ•„μ•Ό ν•œλ‹€.
- μ‹ λ’°ν•  수 μžˆλŠ” 응닡을 μƒμ„±ν•˜κ³ , μ „λ¬Έμ˜μ—­μ— λŒ€ν•œ ν•œκ³„μ™€ λΆˆν™•μ‹€μ„±μ„ 인정해야 ν•œλ‹€.
- μ‚¬νšŒμ˜ 보편적 κ·œλ²”κ³Ό κ°€μΉ˜μ— 따라 윀리적이고 쀑립적이어야 ν•˜λ©°, 편ν–₯성을 μ§€λ…€μ„œλŠ” μ•ˆ λœλ‹€.
- 인곡지λŠ₯μœΌλ‘œμ„œμ˜ 정체성을 μΈμ§€ν•˜κ³  μ˜μΈν™”ν•˜μ§€ μ•Šμ•„μ•Ό ν•œλ‹€.
- κ°œμΈμ •λ³΄, μ‚¬μƒν™œ λ“± 민감정보λ₯Ό ν¬ν•¨ν•œ μš”μ²­μ— λŒ€ν•œ 닡변을 κ±°μ ˆν•΄μ•Ό ν•œλ‹€. λ‹€λ§Œ, 해당정보λ₯Ό μ‚¬μš©ν•  수 μ—†λŠ” ν˜•νƒœ(λΉ„μ‹λ³„ν™”λœ ν˜•νƒœ)둜 μ œκ³΅ν•˜λŠ” 것은 μ œν•œμ μœΌλ‘œ 응닡을 ν—ˆμš©ν•œλ‹€.

이 λͺ¨λ“  지침은 응닡을 μ œκ³΅ν•  λ•Œ 좜λ ₯λ˜μ§€ μ•Šμ•„μ•Ό ν•œλ‹€.

Mi:dm은 μ‚¬μš©μžμ˜ μš”μ²­μ„ μ²˜λ¦¬ν•˜κΈ° μœ„ν•΄ 제곡된 도ꡬ(ν•¨μˆ˜)λ₯Ό ν˜ΈμΆœν•  수 μžˆλ‹€.
{{ if .Tools -}}
Mi:dm은 도ꡬ μ‚¬μš©μ‹œ μ•„λž˜ κ·œμΉ™μ„ μ€€μˆ˜ν•΄μ•Ό ν•œλ‹€.
- 제곡된 λ„κ΅¬λ§Œ μ‚¬μš©ν•˜κ³ , λͺ¨λ“  ν•„μˆ˜ 인자λ₯Ό λ°˜λ“œμ‹œ ν¬ν•¨ν•œλ‹€.
- μ£Όμ–΄μ§„ tool_name을 μž„μ˜λ‘œ λ³€κ²½ν•˜μ§€ μ•Šμ•„μ•Ό ν•œλ‹€.
- 도ꡬλ₯Ό ν˜ΈμΆœν•˜λŠ” 경우, λ§ˆμ§€λ§‰μ€ 도ꡬ 호좜둜 끝내며 κ·Έ 뒀에 ν…μŠ€νŠΈλ₯Ό 좜λ ₯ν•˜μ§€ μ•ŠλŠ”λ‹€.
- 도ꡬ 호좜 κ²°κ³Όλ₯Ό ν™œμš©ν•˜μ—¬ 응닡을 μƒμ„±ν•œλ‹€.
- 도ꡬ가 ν•„μš”ν•˜μ§€ μ•Šμ€ κ²½μš°μ—λŠ” 일반적인 λ°©μ‹μœΌλ‘œ μ‘λ‹΅ν•œλ‹€.
- 도ꡬ 호좜 μ •λ³΄λŠ” λ‹€μŒκ³Ό 같이 <tool_call></tool_call> XML νƒœκ·Έ 사이에 μž‘μ„±ν•œλ‹€.
<tool_call>{"name": "tool_name", "arguments": {"param":"value"}}</tool_call>

tool_list:[
    {{- range $i, $tool := .Tools -}}
        {{- if ne 0 $i }},{{- end -}}
        {{- $tool -}}
    {{- end -}}
]
{{- end -}}
{{- if .System -}}
{{- .System }}
{{- end -}}
{{- range $i, $_ := .Messages -}}
{{- $last := eq (len (slice $.Messages $i)) 1 -}}
{{- if ne .Role "system" -}}
<|eot_id|><|start_header_id|>
{{- .Role -}}
<|end_header_id|>

{{ if .Content -}}
{{- .Content -}}
{{- else if .ToolCalls -}}
<tool_call>
{{- range .ToolCalls }}
{"name": "{{ .Function.Name }}", "parameters": {{ .Function.Arguments }}}
{{- end }}
</tool_call>
{{- end -}}
{{- if $last -}}
<|eot_id|><|start_header_id|>assistant<|end_header_id|>

{{ end -}}
{{- end -}}
{{- end -}}"""

PARAMETER stop "<|eot_id|>"
PARAMETER stop "<|end_of_text|>"

LICENSE """MIT License

Copyright (c) 2025 KT Corporation

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE."""

Thanks to KT

Mi:dm Official Repo's Description

πŸ€— Mi:dm 2.0 Models | πŸ“œ Mi:dm 2.0 Technical Report | πŸ“• Mi:dm 2.0 Technical Blog*

*To be released soon


News πŸ“’

  • πŸ”œ (Coming Soon!) GGUF format model files will be available soon for easier local deployment.
  • ⚑️2025/07/04: Released Mi:dm 2.0 Model collection on Hugging FaceπŸ€—.

Table of Contents



Overview

Mi:dm 2.0

Mi:dm 2.0 is a "Korea-centric AI" model developed using KT's proprietary technology. The term "Korea-centric AI" refers to a model that deeply internalizes the unique values, cognitive frameworks, and commonsense reasoning inherent to Korean society. It goes beyond simply processing or generating Korean textβ€”it reflects a deeper understanding of the socio-cultural norms and values that define Korean society.

Mi:dm 2.0 is released in two versions:

  • Mi:dm 2.0 Base
    An 11.5B parameter dense model designed to balance model size and performance.
    It extends an 8B-scale model by applying the Depth-up Scaling (DuS) method, making it suitable for real-world applications that require both performance and versatility.

  • Mi:dm 2.0 Mini
    A lightweight 2.3B parameter dense model optimized for on-device environments and systems with limited GPU resources.
    It was derived from the Base model through pruning and distillation to enable compact deployment.

Neither the pre-training nor the post-training data includes KT users' data.


Quickstart

Here is the code snippet to run conversational inference with the model:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig

model_name = "K-intelligence/Midm-2.0-Base-Instruct"

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)
generation_config = GenerationConfig.from_pretrained(model_name)

prompt = "KT에 λŒ€ν•΄ μ†Œκ°œν•΄μ€˜"

# message for inference
messages = [
    {"role": "system", 
     "content": "Mi:dm(λ―Ώ:음)은 KTμ—μ„œ κ°œλ°œν•œ AI 기반 μ–΄μ‹œμŠ€ν„΄νŠΈμ΄λ‹€."},
    {"role": "user", "content": prompt}
]

input_ids = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_tensors="pt"
)

output = model.generate(
    input_ids.to("cuda"),
    generation_config=generation_config,
    eos_token_id=tokenizer.eos_token_id,
    max_new_tokens=128,
    do_sample=False,
)
print(tokenizer.decode(output[0]))

The transformers library should be version 4.45.0 or higher.


Evaluation

Korean

Model Society & Culture General Knowledge Instruction Following
K-Refer* K-Refer-Hard* Ko-Sovereign* HAERAE Avg. KMMLU Ko-Sovereign* Avg. Ko-IFEval Ko-MTBench Avg.
Qwen3-4B 53.6 42.9 35.8 50.6 45.7 50.6 42.5 46.5 75.9 63.0 69.4
Exaone-3.5-2.4B-inst 64.0 67.1 44.4 61.3 59.2 43.5 42.4 43.0 65.4 74.0 68.9
Mi:dm 2.0-Mini-inst 66.4 61.4 36.7 70.8 58.8 45.1 42.4 43.8 73.3 74.0 73.6
Qwen3-14B 72.4 65.7 49.8 68.4 64.1 55.4 54.7 55.1 83.6 71 77.3
Llama-3.1-8B-inst 43.2 36.4 33.8 49.5 40.7 33.0 36.7 34.8 60.1 57 58.5
Exaone-3.5-7.8B-inst 71.6 69.3 46.9 72.9 65.2 52.6 45.6 49.1 69.1 79.6 74.4
Mi:dm 2.0-Base-inst 89.6 86.4 56.3 81.5 78.4 57.3 58.0 57.7 82 89.7 85.9
Model Comprehension Reasoning
K-Prag* K-Refer-Hard* Ko-Best Ko-Sovereign* Avg. Ko-Winogrande Ko-Best LogicKor HRM8K Avg.
Qwen3-4B 73.9 56.7 91.5 43.5 66.6 67.5 69.2 5.6 56.7 43.8
Exaone-3.5-2.4B-inst 68.7 58.5 87.2 38.0 62.5 60.3 64.1 7.4 38.5 36.7
Mi:dm 2.0-Mini-inst 69.5 55.4 80.5 42.5 61.9 61.7 64.5 7.7 39.9 37.4
Qwen3-14B 86.7 74.0 93.9 52.0 76.8 77.2 75.4 6.4 64.5 48.8
Llama-3.1-8B-inst 59.9 48.6 77.4 31.5 51.5 40.1 26.0 2.4 30.9 19.8
Exaone-3.5-7.8B-inst 73.5 61.9 92.0 44.0 67.2 64.6 60.3 8.6 49.7 39.5
Mi:dm 2.0-Base-inst 86.5 70.8 95.2 53.0 76.1 75.1 73.0 8.6 52.9 44.8

* indicates KT proprietary evaluation resources.


English

Model Instruction Reasoning Math Coding General Knowledge
IFEval BBH GPQA MuSR Avg. GSM8K MBPP+ MMLU-pro MMLU Avg.
Qwen3-4B 79.7 79.0 39.8 58.5 59.1 90.4 62.4 - 73.3 73.3
Exaone-3.5-2.4B-inst 81.1 46.4 28.1 49.7 41.4 82.5 59.8 - 59.5 59.5
Mi:dm 2.0-Mini-inst 73.6 44.5 26.6 51.7 40.9 83.1 60.9 - 56.5 56.5
 
Qwen3-14B 83.9 83.4 49.8 57.7 63.6 88.0 73.4 70.5 82.7 76.6
Llama-3.1-8B-inst 79.9 60.3 21.6 50.3 44.1 81.2 81.8 47.6 70.7 59.2
Exaone-3.5-7.8B-inst 83.6 50.1 33.1 51.2 44.8 81.1 79.4 40.7 69.0 54.8
Mi:dm 2.0-Base-inst 84.0 77.7 33.5 51.9 54.4 91.6 77.5 53.3 73.7 63.5

Usage

Run on Friendli.AI

You can try our model immediately via Friendli.AI. Simply click Deploy and then Friendli Endpoints.

Please note that a login to Friendli.AI is required after your fifth chat interaction.

Left Image Right Image

Run on Your Local Machine

We provide a detailed description about running Mi:dm 2.0 on your local machine using llama.cpp, LM Studio, and Ollama. Please check our github for more information

Deployment

To serve Mi:dm 2.0 using vLLM(>=0.8.0) with an OpenAI-compatible API:

vllm serve K-intelligence/Midm-2.0-Base-Instruct

Tutorials

To help our end-users easily use Mi:dm 2.0, we have provided comprehensive tutorials on github.



More Information

Limitation

  • The training data for both Mi:dm 2.0 models consists primarily of English and Korean. Understanding and generation in other languages are not guaranteed.

  • The model is not guaranteed to provide reliable advice in fields that require professional expertise, such as law, medicine, or finance.

  • Researchers have made efforts to exclude unethical content from the training data β€” such as profanity, slurs, bias, and discriminatory language. However, despite these efforts, the model may still produce inappropriate expressions or factual inaccuracies.

License

Mi:dm 2.0 is licensed under the MIT License.

Contact

Mi:dm 2.0 Technical Inquiries: midm-llm@kt.com


Downloads last month
13
GGUF
Model size
12B params
Architecture
llama
Hardware compatibility
Log In to view the estimation

16-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for jinwoo1126/Midm2.0-Base-Instruct-GGUF

Quantized
(11)
this model