Instructions to use EdonFetaji/MK-Llama-3.2-1B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use EdonFetaji/MK-Llama-3.2-1B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="EdonFetaji/MK-Llama-3.2-1B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("EdonFetaji/MK-Llama-3.2-1B")
model = AutoModelForCausalLM.from_pretrained("EdonFetaji/MK-Llama-3.2-1B")

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use EdonFetaji/MK-Llama-3.2-1B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "EdonFetaji/MK-Llama-3.2-1B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "EdonFetaji/MK-Llama-3.2-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/EdonFetaji/MK-Llama-3.2-1B

SGLang

How to use EdonFetaji/MK-Llama-3.2-1B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "EdonFetaji/MK-Llama-3.2-1B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "EdonFetaji/MK-Llama-3.2-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "EdonFetaji/MK-Llama-3.2-1B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "EdonFetaji/MK-Llama-3.2-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use EdonFetaji/MK-Llama-3.2-1B with Docker Model Runner:
```
docker model run hf.co/EdonFetaji/MK-Llama-3.2-1B
```

MK Llama-3.2-1B (Continued Pretraining)

Summary

Continued pretraining for Macedonian language on lvstck/macedonian-corpus-cleaned-dedup.
Trained using LoRA adapters on a single A100.

Training

Stage 1: 0.4 epochs on Google Colab A100
Stage 2: 0.9 epochs on A100 - FINKI GPU cluster
Total: 1.3 epochs

Training Metrics

TensorBoard logs are available in the Training metrics tab of this model repository. Logs only available for STAGE 2 The logs include training loss, learning rate, gradient norm, and evaluation metrics .

View the TensorBoard dashboard here: Open TensorBoard training metrics

Notes

This is a continued pretraining checkpoint focused on Macedonian.

Downloads last month: 243

Safetensors

Model size

1B params

Tensor type

BF16

Model tree for EdonFetaji/MK-Llama-3.2-1B

Base model

meta-llama/Llama-3.2-1B

Adapter

(687)

this model

Adapters

1 model

EdonFetaji
/

MK-Llama-3.2-1B