mw-ai/bassie-GPT

mw-ai/ bassie-GPT

5 Downloads Updated 1 year ago

A small testing model based on citations of the famous serie 'Bassie en Adriaan' to learn about finetuning using QLora, source model llama-3.1:8b-instruct

ollama run mw-ai/bassie-GPT

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

from ollama import chat

response = chat(
    model='mw-ai/bassie-GPT',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'mw-ai/bassie-GPT',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

View all →

Name

1 model

Size / Usage

Context

Input

bassie-GPT:latest

4.9GB · 128K context window · Text · 1 year ago

bassie-GPT:latest

4.9GB

128K

Text

Readme

Bassie en Adriaan Citations Model

A small experimental language model fine-tuned on memorable quotes and dialogues from the famous Dutch TV series “Bassie en Adriaan”.
This project serves as a hands-on learning exercise in fine-tuning using QLoRA, with the base model being llama-3.1:8b-instruct.

The primary goals of this model are:

Testing and experimenting with PEFT - QLoRA fine-tuning .
Experimenting with domain-specific language modeling and a sml.
Asking citations about personages from the series without using RAG or adding it to the prompt.

Note: This model is purely for testing and educational purposes, and not intended for production deployment.