Knowledge extraction model for akro.

tools

ollama run akropora/librarian

curl http://localhost:11434/api/chat \
  -d '{
    "model": "akropora/librarian",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='akropora/librarian',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'akropora/librarian',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 2 months ago

2 months ago

2b2498821c8b · 2.5GB

model

archqwen3

parameters4.02B

quantizationQ4_K_M

2.5GB

license

Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR US

11kB

system

You are Librarian, the knowledge extraction model for a local second brain. Your job is to turn raw

659B

params

{ "num_ctx": 16384, "num_predict": 2200, "repeat_penalty": 1.03, "stop": [ "

155B

template

{{- $lastUserIdx := -1 -}} {{- range $idx, $msg := .Messages -}} {{- if eq $msg.Role "user" }}{{ $la

1.4kB

Readme

Librarian

Librarian is Akro’s knowledge extraction model.

It turns raw chats and documents into compact, structured notes for Akro’s long-term memory system.

Librarian is designed to preserve useful context without forcing the main chat model to carry huge context windows.

It focuses on extracting:

user context
important facts
decisions
goals
open items
retrieval keywords
document knowledge

It is intentionally conservative and is instructed not to invent facts, preferences, relationships, or completion states.

Pull

ollama pull akropora/librarian:latest

Use with Akro

Akro expects the local name:

librarian:latest

Create that alias with:

ollama cp akropora/librarian:latest librarian:latest

Librarian is based on Qwen3 4B Instruct and is tuned for structured memory and knowledge extraction.