Instructions to use webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit") prompt = "Write a story about Einstein" messages = [{"role": "user", "content": prompt}] prompt = tokenizer.apply_chat_template( messages, add_generation_prompt=True ) text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit"
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "mlx-lm": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit" } ] } } }Run Pi
# Start Pi in your project directory: pi
- MLX LM
How to use webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Interactive chat REPL mlx_lm.chat --model "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit"
Run an OpenAI-compatible server
# Install MLX LM uv tool install mlx-lm # Start the server mlx_lm.server --model "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit" # Calling the OpenAI-compatible server with curl curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit", "messages": [ {"role": "user", "content": "Hello"} ] }' - Hermes Agent
How to use webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit with Hermes Agent:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit"
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit"
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit" \ --custom-provider-id mlx-lm \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
Ling3-TR-Tiny-Pruned-Hukuk v2 (Apple Silicon MLX 4-bit)
รzel Mรผlkiyet ve Lisans Bildirimi (Proprietary / All Rights Reserved)
LฤฐSANS: proprietary-private (license: other)
Bu model, budanmฤฑล mimarisi, ince ayar aฤฤฑrlฤฑklarฤฑ ve kuantalama dosyalarฤฑ tamamen รzel Mรผlkiyet (Private /
Proprietary) kapsamฤฑndadฤฑr. Tรผm haklarฤฑ saklฤฑdฤฑr (All Rights Reserved). Model sahibinin (webbrain-one) yazฤฑlฤฑ ve
aรงฤฑk izni olmaksฤฑzฤฑn bu modelin veya tรผrevlerinin kopyalanmasฤฑ, daฤฤฑtฤฑlmasฤฑ, รผรงรผncรผ kiลilerle paylaลฤฑlmasฤฑ, ticari veya
aรงฤฑk kaynaklฤฑ ortamlarda yeniden yayฤฑnlanmasฤฑ kesinlikle yasaktฤฑr.
PekBilen masaรผstรผ uygulamasฤฑnฤฑn Apple Silicon Mac'lerde (MLX motoru) kullandฤฑฤฤฑ kรผรงรผk (dรผลรผk bellekli) hukuk modeli.
Aฤฤฑrlฤฑklar webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-GGUF ile aynฤฑdฤฑr.
- Taban:
webbrain-one/ling3-turkish-tiny-pruned(inclusionAI/Ling-3.0-tinyโ Tรผrkรงe budama, 80 uzman/katman, Tรผrkรงe heal; 5,3 milyar parametre, token baลฤฑna ~1,3 milyar aktif). - v2 eฤitimi: Qwen 3.5 9B v2 ile aynฤฑ veri (
webbrain-one/hukuk-dataset-v2, 6.013 รถrnek; uygulamanฤฑn kendi istem biรงimi, her hedef uygulamanฤฑn denetรงilerinden geรงmiล). Uygulama Ling'i dรผลรผnmesiz รงalฤฑลtฤฑrdฤฑฤฤฑ iรงin plan satฤฑrlarฤฑ รงฤฑkarฤฑldฤฑ. Uzman baลฤฑna LoRA (r 32, ฮฑ 32) bรผtรผn doฤrusal katmanlarda, yalnฤฑz asistan token'larฤฑnda kayฤฑp, lr 1e-4, 1 epoch, en รงok 12.288 token.
Dosyalar
| Dosya | Aรงฤฑklama | Boyut |
|---|---|---|
model.safetensors |
4 bit affine, grup 64 (mlx-lm 0.32.2; v1 MLX paketiyle aynฤฑ ayar) |
3,71 GB |
config.json |
model_type: bailing_hybrid, rope_scaling: null (mlx-lm'deki uygulama iรงin) |
|
tokenizer.json, tokenizer_config.json, chat_template.jinja |
taban modelle aynฤฑ |
*.py dosyalarฤฑ yalnฤฑz transformers uyumluluฤu iรงindir; MLX bunlarฤฑ รงalฤฑลtฤฑrmaz.
Kullanฤฑm
pip install "mlx-lm==0.32.2"
from mlx_lm import load, generate
model, tokenizer = load("webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit")
messages = [{"role": "user", "content": "TBK m. 138 uyarฤฑnca aลฤฑrฤฑ ifa gรผรงlรผฤรผnรผn ลartlarฤฑ nelerdir?"}]
prompt = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False, enable_thinking=False)
print(generate(model, tokenizer, prompt=prompt, max_tokens=512))
Uygulama Ling'i dรผลรผnmesiz รงalฤฑลtฤฑrฤฑr (enable_thinking=False); รถrnekleme ayarlarฤฑ sฤฑcaklฤฑk 0,3, top-p 0,9 (Ling v2 bu
ayarlarla deฤerlendirildi).
Deฤerlendirme
รlรงรผmler aynฤฑ aฤฤฑrlฤฑklarฤฑn GGUF sรผrรผmรผyle (IQ4_XS, llama.cpp b11724, dรผลรผnme kapalฤฑ) yapฤฑldฤฑ; MLX paketi uygulama iลlerinde ayrฤฑca รถlรงรผlmedi. Eฤitimde olmayan 20 gรผncel karar รผzerinde uygulamanฤฑn 139 tek รงaฤrฤฑlฤฑk iลi ve 40 mevzuat sorusu; her yanฤฑt uygulamanฤฑn atฤฑf ve gรถrev denetรงileriyle puanlandฤฑ.
| Ling v1 | Ling v2 | Qwen 3.5 9B v2.1 (dรผลรผnme kapalฤฑ) | |
|---|---|---|---|
| Uygulama iลleri | 67 / 139 | 99 | 97 |
| Mevzuat soru-cevap | 36 / 40 | 33 | 38 |
| TRLawBench (97 รSYM sorusu) | 30 | 23 | 47 |
| HukukBERT cloze (Ling yรถntemi: acc / acc_norm / sohbet MCQ) | 50,8 / 49,3 / 62,7 | 40,8 / 42,5 / 56,3 | 73,6 / 71,1 / โ |
v2, uygulamanฤฑn iลlerinde v1'in รงok รถnรผnde ve dรผลรผnme kapalฤฑ Qwen v2.1 dรผzeyinde; v1'in eฤitim verisindeki kapalฤฑ kitap soru-cevaplardan gelen ezber bilgisinin bir kฤฑsmฤฑnฤฑ (cloze, TRLawBench) ise kaybediyor.
Kalite uyarฤฑsฤฑ: Token baลฤฑna ~1,3 milyar aktif parametreli kรผรงรผk bir modeldir; รงฤฑktฤฑlar hukuki gรถrรผล yerine geรงmez ve birincil kaynaktan doฤrulanmalฤฑdฤฑr.
- Downloads last month
- 27
4-bit
Model tree for webbrain-one/ling3-tr-tiny-pruned-hukuk-v2-MLX-4bit
Base model
webbrain-one/ling3-turkish-tiny-pruned