mannix/llava-phi3

mannix/ llava-phi3

2,014 Downloads Updated 2 years ago

A new small LLaVA model fine-tuned from Phi 3 Mini [I-Quants]

vision

ollama run mannix/llava-phi3:iq2_s

curl http://localhost:11434/api/chat \
  -d '{
    "model": "mannix/llava-phi3:iq2_s",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='mannix/llava-phi3:iq2_s',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'mannix/llava-phi3:iq2_s',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

View all →

Name

4 models

Size / Usage

Context

Input

llava-phi3:iq2_s

1.9GB · 4K context window · Text, Image · 2 years ago

llava-phi3:iq2_s

1.9GB

Text, Image

llava-phi3:iq3_xxs

2.1GB · 4K context window · Text, Image · 2 years ago

llava-phi3:iq3_xxs

2.1GB

Text, Image

llava-phi3:iq3_s

2.3GB · 4K context window · Text, Image · 2 years ago

llava-phi3:iq3_s

2.3GB

Text, Image

llava-phi3:iq4_xs

2.7GB · 4K context window · Text, Image · 2 years ago

llava-phi3:iq4_xs

2.7GB

Text, Image

Readme

llava-phi3 is a LLaVA model fine-tuned from Phi 3 Mini 4k, with strong performance benchmarks on par with the original LLaVA model:

References

Hugging Face

GitHub