petkopetkov/Llama3.2-1B-bg-tokenizer

Llama3.2-1B-Bulgarian-tokenizer

• Developed by: petkopetkov
• License: apache-2.0
• Finetuned from model : unsloth/llama-3.2-1b-bnb-4bit

Llama3.2-1B finetuned on datasets translated to Bulgarian language (with tokenizer trained on Bulgarian text):

• MMLU: multiple-choice questions from various branches of knowledge
• Winogrande challenge: testing world knowledge and understanding
• Hellaswag: testing sentence completion
• ARC Easy/Challenge: testing logical reasoning
• GSM-8k: solving multiple-choice questions in high-school mathematics
• MathQA: math word problems

Usage

First, install the Transformers library with:

pip install -U transformers

Run with the pipeline API

import torch
from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="petkopetkov/Llama3.2-1B-bg-tokenizer",
    torch_dtype=torch.bfloat16, 
    device_map="auto"
)

prompt = "Колко е 2 + 2?"

print(pipe(prompt)[0]['generated_text'])