Reasoning model distilled from DeepSeek-R1, enhanced with GRPO using supplementary reasoning datasets.

14b

ollama run dnotitia/dna-r1:14b-fp32

curl http://localhost:11434/api/chat \
  -d '{
    "model": "dnotitia/dna-r1:14b-fp32",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='dnotitia/dna-r1:14b-fp32',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'dnotitia/dna-r1:14b-fp32',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 10 months ago

10 months ago

c8e29f1a1946 · 59GB ·

model

archphi3

parameters14.7B

quantizationF32

59GB

template

{{- if .System }}<|im_start|>system<|im_sep|>{{ .System }}<|im_end|>{{ end }} {{- range $i, $_ := .M

586B

params

{ "num_predict": 4096, "stop": [ "<|im_end|>" ], "temperature": 0.1, "to

83B

Readme

We introduce DNA-R1, a specialized reasoning model optimized for Korean language based on Microsoft’s Phi-4. By applying large-scale reinforcement learning (RL) using the same methodology as DeepSeek-R1, we have significantly enhanced the model’s Korean reasoning capabilities. This model demonstrates deep understanding of Korean text and exhibits exceptional reasoning abilities across mathematics, coding, and general reasoning tasks.

References

Hugging Face