Nac/smolvlm

Nac/ smolvlm:latest

24 Downloads Updated 3 weeks ago

A deterministic configuration of SmolVLM.

vision

ollama run Nac/smolvlm

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

from ollama import chat

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

import ollama from 'ollama'

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

Details

Updated 3 weeks ago

3 weeks ago

a38697e9d57d · 546MB ·

model

archllama

parameters409M

quantizationQ8_0

437MB

projector

archclip

parameters98.2M

quantizationQ8_0

109MB

template

{{ .Prompt }} ## Image Description

35B

params

{ "stop": [ "<|im_end|>", "\n\n---", "\n\n#", "\n\n**Chain",

164B

Readme

This is a deterministic configuration of SmolVLM made to make image descriptions. Call the LLM with a newline and one image only like this:

curl http://localhost:11434/api/generate -d '{
  "model": "Nac/smolvlm",
  "prompt": "\n",
  "images": ["<base64_encoded_image>"]
}'

Ported from ggml-org/SmolVLM-500M-Instruct-GGUF.