A medically fine-tuned model by BlueishAI.

ollama run blueishai/medic-1.5b-firstaid:q8

curl http://localhost:11434/api/chat \
  -d '{
    "model": "blueishai/medic-1.5b-firstaid:q8",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='blueishai/medic-1.5b-firstaid:q8',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'blueishai/medic-1.5b-firstaid:q8',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 3 weeks ago

3 weeks ago

1b80f5f028c3 · 1.9GB ·

model

archqwen2

parameters1.78B

quantizationQ8_0

1.9GB

template

{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user

156B

system

You are Medic, a tiny first-aid assistant. Give fast, practical, safe first-aid steps. Tell the user

178B

params

{ "num_ctx": 512, "num_predict": 96, "repeat_penalty": 1.08, "stop": [ "<|im

143B

Readme

Medic

Medic is a medically aimed expertise model by BlueishAI. It aims to be lightweight model that is trained on medical datasets and has knowledge about first-aid and the field of medicines.

Available Variants

File	Quantization
`medic-1.5b-firstaid-Q4_K_M.gguf`	Q4_K_M
`medic-1.5b-firstaid-Q8_0.gguf`	Q8_0

Architecture

The medic model was fine-tuned from anant-base-1.5b which is QLORA adoption from qwen2.5. Altogether, its a model based on qwen. The training method used was LoRA supervised fine-tuning and the datasets used were i-am-mushfiq/FirstAidQA.

Usage

The model is fully open-source, including its source code and model weights. Anyone can use this and anyone can use this. It was aimed to be efficient and helpful at best cases.