Instructions to use stukenov/sozkz-core-llama-300m-kk-gec-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use stukenov/sozkz-core-llama-300m-kk-gec-v3 with Transformers:

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

pipe = pipeline("text-generation", model="stukenov/sozkz-core-llama-300m-kk-gec-v3")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("stukenov/sozkz-core-llama-300m-kk-gec-v3")
model = AutoModelForMultimodalLM.from_pretrained("stukenov/sozkz-core-llama-300m-kk-gec-v3")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use stukenov/sozkz-core-llama-300m-kk-gec-v3 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-300m-kk-gec-v3"
# 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-300m-kk-gec-v3",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/stukenov/sozkz-core-llama-300m-kk-gec-v3

SGLang

How to use stukenov/sozkz-core-llama-300m-kk-gec-v3 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-300m-kk-gec-v3" \
    --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-300m-kk-gec-v3",
		"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-300m-kk-gec-v3" \
        --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-300m-kk-gec-v3",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use stukenov/sozkz-core-llama-300m-kk-gec-v3 with Docker Model Runner:
```
docker model run hf.co/stukenov/sozkz-core-llama-300m-kk-gec-v3
```

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.

YAML Metadata Warning:The pipeline tag "text2text-generation" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

SozKZ Core Llama 300M — Kazakh GEC v3

Казақ тіліне арналған грамматикалық қателерді түзету моделі (Grammar Error Correction).

Model Details

Parameter	Value
Architecture	LlamaForCausalLM (decoder-only)
Parameters	325M
Base model	`stukenov/sozkz-core-llama-300m-kk-base-v1`
Training data	sozkz-corpus-synthetic-kk-gec-v1
Training	5 epochs, LR=2e-5, BS=128, cosine schedule
Clean ratio	50% (error pairs + identity pairs)
Hardware	4× RTX 4090 (vast.ai), 48 min
License	MIT

Format

The model uses a tag-based correction format:

<тег> ошибочный текст
→ исправленный текст

Available Tags

Tag	Error Type	Example
`<грамматика>`	General grammar	catch-all
`<сингармонизм>`	Vowel harmony	әкімшілігінда → әкімшілігінде
`<септік>`	Case suffixes	гранитоидтерке → гранитоидтерге
`<тәуелдік>`	Possessive	түпкы → түпкі
`<жіктік>`	Personal endings	тарасың → тарады
`<шылау>`	Postpositions	білікге үшін → білік үшін
`<көптік>`	Plural	Студенттар → Студенттер
`<болымсыз>`	Negation	жарамеған → жарамаған
`<шақ>`	Tense	кетееді → кетеді
`<жалғау>`	Suffixes	general suffix errors
`<қате>`	Typos/noise	Астна → Астана
`<таза>`	Clean (no error)	should return input unchanged

Usage

from transformers import AutoModelForCausalLM, GPT2TokenizerFast
from huggingface_hub import hf_hub_download
import torch

model_id = "stukenov/sozkz-core-llama-300m-kk-gec-v3"
tok_file = hf_hub_download(repo_id=model_id, filename="tokenizer.json")
tokenizer = GPT2TokenizerFast(tokenizer_file=tok_file)
tokenizer.pad_token = tokenizer.eos_token

model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16)
model.eval()

def correct(tag, text):
    prompt = f"<{tag}> {text}\n→ "
    inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=480)
    with torch.no_grad():
        out = model.generate(
            **inputs,
            max_new_tokens=len(inputs["input_ids"][0]) + 30,
            temperature=0.3, top_p=0.9, do_sample=True,
            repetition_penalty=1.1,
            pad_token_id=tokenizer.eos_token_id,
        )
    result = tokenizer.decode(out[0], skip_special_tokens=True)
    if "→ " in result:
        result = result.split("→ ", 1)[1]
    for stop in ["\n<", "\n\n"]:
        if stop in result:
            result = result[:result.index(stop)]
    return result.strip() or text

# Example
print(correct("көптік", "Студенттар университетте оқиды."))
# → Студенттер университетте оқиды.

Evaluation (v3 vs previous versions)

Metric	v1 (1ep)	v2a (1ep)	v3 (5ep)
Accuracy	9.0%	6.6%	15.8%
Wrong	85.2%	78.4%	70.8%
FP (clean)	84%	82%	76%

Best per-tag accuracy: көптік 32%, болымсыз 30%, жіктік 26%.

Limitations

Accuracy is still low (~16%) — this is an early research model
False positive rate is high (76%) — the model sometimes changes correct text
Occasional non-Kazakh output (Chinese/Greek characters from base Llama)
Best suited for research, not production use

Part of SozKZ Project

This model is part of the SozKZ project for Kazakh language AI.

Citation

@misc{sozkz-gec-v3-2026,
  title={SozKZ Core Llama 300M Kazakh GEC v3},
  author={Stukenov, Saken},
  year={2026},
  url={https://huggingface.co/stukenov/sozkz-core-llama-300m-kk-gec-v3}
}

Benchmark Results

Evaluated on 100-example custom GEC test (pure model inference, no pre/post pipeline).

Category	Score
Орфография (емле)	0/30 (0%)
Грамматика	1/20 (5%)
Пунктуация	0/15 (0%)
Смешанный	0/20 (0%)
Identity preservation	0/15 (0%)
Total	1/100 (1%)

Leaderboard (100-example custom benchmark)

Модель	Total	Емле/30	Грамм/20	Пункт/15	Смеш/20	Ident/15
sozkz-core-llama-600m-kk-gec-v1	47%	15	12	3	2	15/15
sozkz-fix-qwen-500m-kk-gec-v3	38%	0	16	9	0	13/15
sozkz-core-llama-300m-kk-gec-v4	37%	9	6	4	3	15/15
sozkz-fix-qwen-500m-kk-gec-v1	35%	0	12	8	0	15/15
sozkz-fix-qwen-500m-kk-gec-v2	30%	0	11	7	0	12/15
sozkz-core-llama-1b-kk-gec-v1	16%	2	6	1	0	7/15
sozkz-fix-qwen-500m-kk-gec-v4	5%	0	1	4	0	0/15
sozkz-fix-mt5b-kk-gec-run13-v1	5%	0	2	0	0	3/15
sozkz-nllb-1b-kk-gec-v1	1%	0	1	0	0	0/15
sozkz-nllb-1b-kk-pretrain-v1	1%	0	1	0	0	0/15
sozkz-core-llama-300m-kk-gec-v3	1%	0	1	0	0	0/15
sozkz-core-llama-300m-kk-gec-v1/v2a/v2b	0–1%	0	0	0	0	0–1
sozkz-fix-mt5-50m-kk-gec-v1	0%	0	0	0	0	0/15

Downloads last month: -

Safetensors

Model size

0.3B params

Tensor type

BF16

Collection including stukenov/sozkz-core-llama-300m-kk-gec-v3

SozKZ GEC: Kazakh Grammar Error Correction

Collection

Grammar error correction models and datasets for Kazakh — Llama GEC (300M, 600M), mT5 GEC, morphology models • 10 items • Updated Apr 7