Instructions to use stukenov/sozkz-core-llama-50m-kk-balanced-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use stukenov/sozkz-core-llama-50m-kk-balanced-v1 with Transformers:

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

pipe = pipeline("text-generation", model="stukenov/sozkz-core-llama-50m-kk-balanced-v1")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("stukenov/sozkz-core-llama-50m-kk-balanced-v1")
model = AutoModelForCausalLM.from_pretrained("stukenov/sozkz-core-llama-50m-kk-balanced-v1")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use stukenov/sozkz-core-llama-50m-kk-balanced-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "stukenov/sozkz-core-llama-50m-kk-balanced-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "stukenov/sozkz-core-llama-50m-kk-balanced-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/stukenov/sozkz-core-llama-50m-kk-balanced-v1

SGLang

How to use stukenov/sozkz-core-llama-50m-kk-balanced-v1 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 "stukenov/sozkz-core-llama-50m-kk-balanced-v1" \
    --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": "stukenov/sozkz-core-llama-50m-kk-balanced-v1",
		"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 "stukenov/sozkz-core-llama-50m-kk-balanced-v1" \
        --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": "stukenov/sozkz-core-llama-50m-kk-balanced-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use stukenov/sozkz-core-llama-50m-kk-balanced-v1 with Docker Model Runner:
```
docker model run hf.co/stukenov/sozkz-core-llama-50m-kk-balanced-v1
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Kazakh LLaMA 50M — Balanced (Soz)

A LLaMA-architecture language model with ~50M parameters trained on a domain-balanced Kazakh corpus.

Overview

Property	Value
Parameters	~50M
Architecture	LLaMA (RoPE, SwiGLU, RMSNorm)
Training data	Domain-balanced Kazakh corpus
License	Apache 2.0

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("stukenov/kazakh-gpt2-50k")
model = AutoModelForCausalLM.from_pretrained("stukenov/kazakh-llama-50m-balanced")

input_ids = tokenizer("Қазақстан — ", return_tensors="pt").input_ids
output = model.generate(input_ids, max_new_tokens=50, do_sample=True, temperature=0.8)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Training Details

Trained from scratch on a balanced multi-domain Kazakh corpus
Precision: bfloat16

Project

Part of the Soz — Kazakh Language Models project, a research effort to build open-source language models for Kazakh.

Citation

@misc{tukenov2026soz,
  title={Soz: Small Language Models for Kazakh},
  author={Tukenov, Saken},
  year={2026},
  url={https://huggingface.co/stukenov/kazakh-llama-50m-balanced}
}

License

Apache 2.0

Downloads last month: -

Safetensors

Model size

50.3M params

Tensor type

F32

Collection including stukenov/sozkz-core-llama-50m-kk-balanced-v1

SozKZ Core: Kazakh Language Models

Collection

Base, instruct, and balanced Kazakh language models trained from scratch — Llama (50M–600M), GPT2, Pythia architectures • 22 items • Updated Mar 25