{"version":1,"pages":[{"id":"SilL6NK0Ks0r3qBzCYAf","title":"Unsloth Docs","pathname":"/docs","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9a5","description":"Unsloth is an open-source framework for running and training LLMs.","breadcrumbs":[{"label":"Get Started"}]},{"id":"Du8yzXwHIEJYlMtjjyGw","title":"Fine-tuning for Beginners","pathname":"/docs/get-started/fine-tuning-for-beginners","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"2b50","description":"","breadcrumbs":[{"label":"Get Started"}]},{"id":"odJXZM9jv284RKqZ2Pna","title":"Unsloth Requirements","pathname":"/docs/get-started/fine-tuning-for-beginners/unsloth-requirements","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f6e0","description":"Here are Unsloth's requirements including system and GPU VRAM requirements.","breadcrumbs":[{"label":"Get Started"},{"label":"Fine-tuning for Beginners","emoji":"2b50"}]},{"id":"HP82bIzgldwxWk3OSzVy","title":"FAQ + Is Fine-tuning Right For Me?","pathname":"/docs/get-started/fine-tuning-for-beginners/faq-+-is-fine-tuning-right-for-me","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f914","description":"If you're stuck on if fine-tuning is right for you, see here! Learn about fine-tuning misconceptions, how it compared to RAG and more:","breadcrumbs":[{"label":"Get Started"},{"label":"Fine-tuning for Beginners","emoji":"2b50"}]},{"id":"bISOEydFwcVt8cnfyCfS","title":"Unsloth Notebooks","pathname":"/docs/get-started/unsloth-notebooks","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4d2","description":"Fine-tuning notebooks: Explore the Unsloth catalog.","breadcrumbs":[{"label":"Get Started"}]},{"id":"nQlzs5BcvqlaEjhsgbtY","title":"Unsloth Model Catalog","pathname":"/docs/get-started/unsloth-model-catalog","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f52e","description":"","breadcrumbs":[{"label":"Get Started"}]},{"id":"WbSfE0ITQYsNqERZwnbZ","title":"Unsloth Installation","pathname":"/docs/get-started/install","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4e5","description":"Learn to install Unsloth locally or online.","breadcrumbs":[{"label":"Get Started"}]},{"id":"LhZlVJv6yLKmWbNy4UmP","title":"Install Unsloth via pip and uv","pathname":"/docs/get-started/install/pip-install","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"desktop-arrow-down","description":"To install Unsloth locally via Pip, follow the steps below:","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"nQJglux1e9VKfVL4F43M","title":"Install Unsloth on MacOS","pathname":"/docs/get-started/install/mac","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"apple","description":"","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"Sv0QKHkAGvwTK47OKX2A","title":"How to Fine-Tune LLMs on Windows with Unsloth (Step-by-Step Guide)","pathname":"/docs/get-started/install/windows-installation","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"windows","description":"See how to install Unsloth on Windows to start fine-tuning LLMs locally.","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"dZaYUyA34oYX3LotyGAB","title":"Install Unsloth via Docker","pathname":"/docs/get-started/install/docker","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"docker","description":"Install Unsloth using our official Docker container","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"SQoCZEpSeGsxtIypEgup","title":"Updating Unsloth","pathname":"/docs/get-started/install/updating","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"arrow-rotate-right","description":"To update or use an old version of Unsloth, follow the steps below:","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"GUxDmG8LaAiQinraCXCr","title":"Fine-tuning LLMs on AMD GPUs with Unsloth Guide","pathname":"/docs/get-started/install/amd","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"square-up-right","description":"Learn how to fine-tune large language models (LLMs) on AMD GPUs with Unsloth.","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"FhVmcV9yU5zmKQvNYNb8","title":"Fine-tuning LLMs on Intel GPUs with Unsloth","pathname":"/docs/get-started/install/intel","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"info","description":"Learn how to train and fine-tune large language models on Intel GPUs.","breadcrumbs":[{"label":"Get Started"},{"label":"Unsloth Installation","emoji":"1f4e5"}]},{"id":"nw2c1elNySGBBav8WP9B","title":"Fine-tuning LLMs Guide","pathname":"/docs/get-started/fine-tuning-llms-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9ec","description":"Learn all the basics and best practices of fine-tuning. Beginner-friendly.","breadcrumbs":[{"label":"Get Started"}]},{"id":"XgcpRfamZHmHnRHBnVE4","title":"Datasets Guide","pathname":"/docs/get-started/fine-tuning-llms-guide/datasets-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4c8","description":"Learn how to create & prepare a dataset for fine-tuning.","breadcrumbs":[{"label":"Get Started"},{"label":"Fine-tuning LLMs Guide","emoji":"1f9ec"}]},{"id":"y6obKRSk8TwyjIrCjuGE","title":"LoRA fine-tuning Hyperparameters Guide","pathname":"/docs/get-started/fine-tuning-llms-guide/lora-hyperparameters-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9e0","description":"Learn step-by-step the best LLM fine-tuning settings - LoRA rank & alpha, epochs, batch size + gradient accumulation, QLoRA vs. LoRA, target modules, and more.","breadcrumbs":[{"label":"Get Started"},{"label":"Fine-tuning LLMs Guide","emoji":"1f9ec"}]},{"id":"BSShKhLoFNlGWO5cN8VJ","title":"What Model Should I Use for Fine-tuning?","pathname":"/docs/get-started/fine-tuning-llms-guide/what-model-should-i-use","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"2753","description":"","breadcrumbs":[{"label":"Get Started"},{"label":"Fine-tuning LLMs Guide","emoji":"1f9ec"}]},{"id":"cECKVbf1TpF5j7WC0riJ","title":"Tutorial: How to Finetune Llama-3 and Use In Ollama","pathname":"/docs/get-started/fine-tuning-llms-guide/tutorial-how-to-finetune-llama-3-and-use-in-ollama","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f999","description":"Beginner's Guide for creating a customized personal assistant (like ChatGPT) to run locally on Ollama","breadcrumbs":[{"label":"Get Started"},{"label":"Fine-tuning LLMs Guide","emoji":"1f9ec"}]},{"id":"vT6jTKG1LCfN7HoJ4fVR","title":"Reinforcement Learning (RL) Guide","pathname":"/docs/get-started/reinforcement-learning-rl-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4a1","description":"Learn all about Reinforcement Learning (RL) and how to train your own DeepSeek-R1 reasoning model with Unsloth using GRPO. A complete guide from beginner to advanced.","breadcrumbs":[{"label":"Get Started"}]},{"id":"JdGEInjVZuXH42Y6O2Kz","title":"Reinforcement Learning GRPO with 7x Longer Context","pathname":"/docs/get-started/reinforcement-learning-rl-guide/grpo-long-context","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f300","description":"Learn how Unsloth enables ultra long context RL fine-tuning.","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"aV6S9cmDmSv5ky4ZCo5d","title":"Vision Reinforcement Learning (VLM RL)","pathname":"/docs/get-started/reinforcement-learning-rl-guide/vision-reinforcement-learning-vlm-rl","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f441-1f5e8","description":"Train Vision/multimodal models via GRPO and RL with Unsloth!","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"pFyRT83vVFiXANdxamCs","title":"FP8 Reinforcement Learning","pathname":"/docs/get-started/reinforcement-learning-rl-guide/fp8-reinforcement-learning","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f3b1","description":"Train reinforcement learning (RL) and GRPO in FP8 precision with Unsloth.","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"QbifW3DsTPYvRJcVkXYz","title":"Tutorial: Train your own Reasoning model with GRPO","pathname":"/docs/get-started/reinforcement-learning-rl-guide/tutorial-train-your-own-reasoning-model-with-grpo","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"26a1","description":"Beginner's Guide to transforming a model like Llama 3.1 (8B) into a reasoning model by using Unsloth and GRPO.","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"P9PfsQ0BjnZuwPaXA93E","title":"Advanced Reinforcement Learning Documentation","pathname":"/docs/get-started/reinforcement-learning-rl-guide/advanced-rl-documentation","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9e9","description":"Advanced documentation settings when using Unsloth with GRPO.","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"FoDSrxpEqxCs9VqqDJpp","title":"GSPO Reinforcement Learning","pathname":"/docs/get-started/reinforcement-learning-rl-guide/advanced-rl-documentation/gspo-reinforcement-learning","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"lightbulb-on","description":"Train with GSPO (Group Sequence Policy Optimization) RL in Unsloth.","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"},{"label":"Advanced Reinforcement Learning Documentation","emoji":"1f9e9"}]},{"id":"PVxTK4Eal3B77LmfhdRU","title":"RL Reward Hacking","pathname":"/docs/get-started/reinforcement-learning-rl-guide/advanced-rl-documentation/rl-reward-hacking","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"treasure-chest","description":"Learn what is Reward Hacking in Reinforcement Learning and how to counter it.","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"},{"label":"Advanced Reinforcement Learning Documentation","emoji":"1f9e9"}]},{"id":"wQboNZf1ZtBJ9Qk4WQxb","title":"FP16 vs BF16 for RL","pathname":"/docs/get-started/reinforcement-learning-rl-guide/advanced-rl-documentation/fp16-vs-bf16-for-rl","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"2049","description":"Defeating the Training-Inference Mismatch via FP16 https://arxiv.org/pdf/2510.26788 shows how using float16 is better than bfloat16","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"},{"label":"Advanced Reinforcement Learning Documentation","emoji":"1f9e9"}]},{"id":"VbqKpm1bqUCSPsLzKkhe","title":"Memory Efficient RL","pathname":"/docs/get-started/reinforcement-learning-rl-guide/memory-efficient-rl","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"memory","description":"","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"dJcI8cUqSvuD1WuVj6qU","title":"Preference Optimization Training - DPO, ORPO & KTO","pathname":"/docs/get-started/reinforcement-learning-rl-guide/preference-dpo-orpo-and-kto","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f3c6","description":"Learn about preference alignment fine-tuning with DPO, GRPO, ORPO or KTO via Unsloth, follow the steps below:","breadcrumbs":[{"label":"Get Started"},{"label":"Reinforcement Learning (RL) Guide","emoji":"1f4a1"}]},{"id":"qyazJc8QbOQ0mtlu6uEv","title":"Introducing Unsloth Studio","pathname":"/docs/new/studio","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9a5","description":"Run and train AI models locally with Unsloth Studio.","breadcrumbs":[{"label":"New"}]},{"id":"vrLQd9559vRkDY8zRR0h","title":"Get started with Unsloth Studio","pathname":"/docs/new/studio/start","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"bolt","description":"A guide for getting started with the fine-tuning studio, data recipes, model exporting, and chat.","breadcrumbs":[{"label":"New"},{"label":"Introducing Unsloth Studio","emoji":"1f9a5"}]},{"id":"FdMvLj95MbkAR4aHURvS","title":"How to Run models with Unsloth Studio","pathname":"/docs/new/studio/chat","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"comment-dots","description":"Run AI models, LLMs and GGUFs locally with Unsloth Studio.","breadcrumbs":[{"label":"New"},{"label":"Introducing Unsloth Studio","emoji":"1f9a5"}]},{"id":"XFZRr9F9hSOSIbG5lxqB","title":"Unsloth Studio Installation","pathname":"/docs/new/studio/install","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"arrow-down-to-square","description":"Learn how to install Unsloth Studio on your local device.","breadcrumbs":[{"label":"New"},{"label":"Introducing Unsloth Studio","emoji":"1f9a5"}]},{"id":"m9k4PLFmjpsAP6LsQt7u","title":"Unsloth Data Recipes","pathname":"/docs/new/studio/data-recipe","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"hat-chef","description":"Learn how to create, build and edit datasets with Unsloth Studio's Data Recipes.","breadcrumbs":[{"label":"New"},{"label":"Introducing Unsloth Studio","emoji":"1f9a5"}]},{"id":"5ZU2kPF2eJ7VK0GeEUhu","title":"Export models with Unsloth Studio","pathname":"/docs/new/studio/export","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"box-isometric","description":"Learn how to export your safetensor or LoRA model files to GGUF or other formats.","breadcrumbs":[{"label":"New"},{"label":"Introducing Unsloth Studio","emoji":"1f9a5"}]},{"id":"OktH76Rsg2WQ12B4KR5H","title":"Unsloth Updates","pathname":"/docs/new/changelog","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"sparkles","description":"Unsloth Changelog for our latest releases, improvements and fixes.","breadcrumbs":[{"label":"New"}]},{"id":"NpuhjPsxi8BKhuS8nnyY","title":"Qwen3.6 - How to Run Locally","pathname":"/docs/models/qwen3.6","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f49c","description":"Run the new Qwen3.6-27B and 35B-A3B models locally!","breadcrumbs":[{"label":"Models"}]},{"id":"VnmWq1kNppQrTqCI6aLH","title":"Gemma 4 - How to Run Locally","pathname":"/docs/models/gemma-4","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"2728","description":"Run Google’s new Gemma 4 models locally, including E2B, E4B, 26B A4B, and 31B.","breadcrumbs":[{"label":"Models"}]},{"id":"6iXghkDoe3jzknTq5aWx","title":"Gemma 4 Fine-tuning Guide","pathname":"/docs/models/gemma-4/train","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"flask-gear","description":"Train Gemma 4 by Google with Unsloth.","breadcrumbs":[{"label":"Models"},{"label":"Gemma 4 - How to Run Locally","emoji":"2728"}]},{"id":"JcwJOcoquFknfeDFxM7k","title":"Qwen3.5 - How to Run Locally","pathname":"/docs/models/qwen3.5","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f49c","description":"Run the new Qwen3.5 LLMs including Medium: Qwen3.5-35B-A3B, 27B, 122B-A10B, Small: Qwen3.5-0.8B, 2B, 4B, 9B and 397B-A17B on your local device!","breadcrumbs":[{"label":"Models"}]},{"id":"PzWNOGBEqMa4Xa1reosr","title":"Qwen3.5 Fine-tuning Guide","pathname":"/docs/models/qwen3.5/fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"flask-gear","description":"Learn how to fine-tune Qwen3.5 LLMs with Unsloth.","breadcrumbs":[{"label":"Models"},{"label":"Qwen3.5 - How to Run Locally","emoji":"1f49c"}]},{"id":"jb9Bhr7e6quGvUmcIcZe","title":"Qwen3.5 GGUF Benchmarks","pathname":"/docs/models/qwen3.5/gguf-benchmarks","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"chart-fft","description":"See how Unsloth Dynamic GGUFs perform + analysis of perplexity, KL divergence & MXFP4.","breadcrumbs":[{"label":"Models"},{"label":"Qwen3.5 - How to Run Locally","emoji":"1f49c"}]},{"id":"ftHgogOloVhCFawwwmFL","title":"Kimi K2.6 - How to Run Locally","pathname":"/docs/models/kimi-k2.6","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f95d","description":"Step-by-step guide to running Kimi-K2.6 on your own local device.","breadcrumbs":[{"label":"Models"}]},{"id":"GEABCCTb5KV7QKOeL4YY","title":"NVIDIA Nemotron 3 Nano Omni - How To Run Locally","pathname":"/docs/models/nemotron-3-nano-omni","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9e9","description":"Run & fine-tune Nemotron-3-Nano-Omni-30B-A3B locally on your device!","breadcrumbs":[{"label":"Models"}]},{"id":"BAeSP6aOxvSeDUzCgKOK","title":"Large language model (LLMs) Tutorials","pathname":"/docs/models/tutorials","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f680","description":"","breadcrumbs":[{"label":"Models"}]},{"id":"Omr0gGekk3zqhZtwQdmd","title":"Qwen3 - How to Run & Fine-tune","pathname":"/docs/models/tutorials/qwen3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f320","description":"Learn to run & fine-tune Qwen3 locally with Unsloth + our Dynamic 2.0 quants","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"DM6MVuVKY1zKR9oL23wQ","title":"Qwen3-VL: How to Run Guide","pathname":"/docs/models/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-vl-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f320","description":"Learn to fine-tune and run Qwen3-VL locally with Unsloth.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"},{"label":"Qwen3 - How to Run & Fine-tune","emoji":"1f320"}]},{"id":"o5rHE4o7g4QZc09TMMpj","title":"Qwen3-2507: Run Locally Guide","pathname":"/docs/models/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-2507","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f320","description":"Run Qwen3-30B-A3B-2507 and 235B-A22B Thinking and Instruct versions locally on your device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"},{"label":"Qwen3 - How to Run & Fine-tune","emoji":"1f320"}]},{"id":"vQSH3015yDj4lD5ssFIL","title":"MiniMax-M2.7 - How to Run Locally","pathname":"/docs/models/tutorials/minimax-m27","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"waveform","description":"Run MiniMax-M2.7 LLM locally on your own device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"euF2aUT7116RmwwAtw3R","title":"GLM-5: How to Run Locally Guide","pathname":"/docs/models/tutorials/glm-5","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"z","description":"Run the new GLM-5 model by Z.ai on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"pPNg7FOowDGW06Ii95o6","title":"Kimi K2.5: How to Run Locally Guide","pathname":"/docs/models/tutorials/kimi-k2.5","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f95d","description":"Guide on running Kimi-K2.5 on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"zWg6fYWpGDwkPXd58ReM","title":"GLM-4.7-Flash: How To Run Locally","pathname":"/docs/models/tutorials/glm-4.7-flash","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"z","description":"Run & fine-tune GLM-4.7-Flash locally on your device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"VgFqcc02RzjCTHLApvjO","title":"MiniMax-M2.5: How to Run Guide","pathname":"/docs/models/tutorials/minimax-m25","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"waveform","description":"Run MiniMax-M2.5 locally on your own device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"rxJa5vhJrnBcNUFsBLsJ","title":"Qwen3-Coder: How to Run Locally","pathname":"/docs/models/tutorials/qwen3-coder-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f320","description":"Run Qwen3-Coder-30B-A3B-Instruct and 480B-A35B locally with Unsloth Dynamic quants.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"U0cxXh4tBSRb2I1IZbW5","title":"Gemma 3 - How to Run Guide","pathname":"/docs/models/tutorials/gemma-3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"google","description":"How to run Gemma 3 effectively with our GGUFs on llama.cpp, Ollama, Open WebUI and how to fine-tune with Unsloth!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"TkprHPXbQktLaSKR2A3L","title":"Gemma 3n: How to Run & Fine-tune","pathname":"/docs/models/tutorials/gemma-3-how-to-run-and-fine-tune/gemma-3n-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"google","description":"Run Google's new Gemma 3n locally with Dynamic GGUFs on llama.cpp, Ollama, Open WebUI and fine-tune with Unsloth!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"},{"label":"Gemma 3 - How to Run Guide","icon":"google"}]},{"id":"aQX8YMqzttGdCG0oWaHQ","title":"DeepSeek-OCR 2: How to Run & Fine-tune Guide","pathname":"/docs/models/tutorials/deepseek-ocr-2","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f433","description":"Guide on how to run and fine-tune DeepSeek-OCR-2 locally.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"1wSv7BAqW26rePIFITZn","title":"GLM-4.7: How to Run Locally Guide","pathname":"/docs/models/tutorials/glm-4.7","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"z","description":"A guide on how to run Z.ai GLM-4.7 model on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"2jzPq5XdOzNHmR7wDnnQ","title":"How to Run Qwen-Image-2512 Locally in ComfyUI","pathname":"/docs/models/tutorials/qwen-image-2512","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f49f","description":"Step-by-step tutorial for running Qwen-Image-2512 on your local device with ComfyUI.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"sUIzzNUN3nVoZ7bi99AS","title":"Run Qwen-Image-2512 in stable-diffusion.cpp Tutorial","pathname":"/docs/models/tutorials/qwen-image-2512/stable-diffusion.cpp","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f3a8","description":"Tutorial for using Qwen-Image-2512 in stable-diffusion.cpp.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"},{"label":"How to Run Qwen-Image-2512 Locally in ComfyUI","emoji":"1f49f"}]},{"id":"3Liotgx1T4MeF1J8CO0m","title":"Devstral 2 - How to Run Guide","pathname":"/docs/models/tutorials/devstral-2","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4d9","description":"Guide for local running Mistral Devstral 2 models: 123B-Instruct-2512 and Small-2-24B-Instruct-2512.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"zRbjQXuLmfdZD90U410W","title":"Ministral 3 - How to Run Guide","pathname":"/docs/models/tutorials/ministral-3","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f431","description":"Guide for Mistral Ministral 3 models, to run or fine-tune locally on your device","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"ckxEaylNcpFEtJmmgkoU","title":"DeepSeek-OCR: How to Run & Fine-tune","pathname":"/docs/models/tutorials/deepseek-ocr-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f433","description":"Guide on how to run and fine-tune DeepSeek-OCR locally.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"fLpotmQAWm06ZmVkliIj","title":"Kimi K2 Thinking: Run Locally Guide","pathname":"/docs/models/tutorials/kimi-k2-thinking-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f319","description":"Guide on running Kimi-K2-Thinking and Kimi-K2 on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"kubJWq6dZSW06gdjy3QO","title":"GLM-4.6: Run Locally Guide","pathname":"/docs/models/tutorials/glm-4.6-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"z","description":"A guide on how to run Z.ai GLM-4.6 and GLM-4.6V-Flash model on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"cUiTofDNgkP12VQLa9cl","title":"Qwen3-Next: Run Locally Guide","pathname":"/docs/models/tutorials/qwen3-next","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f320","description":"Run Qwen3-Next-80B-A3B-Instruct and Thinking versions locally on your device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"J4GtBjM0f3vuIwoSfnHy","title":"FunctionGemma: How to Run & Fine-tune","pathname":"/docs/models/tutorials/functiongemma","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"google","description":"Learn how to run and fine-tune FunctionGemma locally on your device and phone.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"eK0BfjMHNrfvfe4HI6Pk","title":"DeepSeek-V3.1: How to Run Locally","pathname":"/docs/models/tutorials/deepseek-v3.1-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f40b","description":"A guide on how to run DeepSeek-V3.1 and Terminus on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"Ji5D1Y22Eu0ZtCpXU9ja","title":"DeepSeek-R1-0528: How to Run Locally","pathname":"/docs/models/tutorials/deepseek-r1-0528-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f40b","description":"A guide on how to run DeepSeek-R1-0528 including Qwen3 on your own local device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"CixcWjfA7MpVVdwbzTmg","title":"Liquid LFM2.5: How To Run & Fine-tune","pathname":"/docs/models/tutorials/lfm2.5","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4a7","description":"Run and fine-tune LFM2.5 Instruct and Vision locally on your device!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"WYCU1j3i1h4Z2y63PJn9","title":"Magistral: How to Run & Fine-tune","pathname":"/docs/models/tutorials/magistral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4a5","description":"Meet Magistral - Mistral's new reasoning models.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"A6Kc1s4GOsHXR07kaEqe","title":"IBM Granite 4.0","pathname":"/docs/models/tutorials/ibm-granite-4.0","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"cube","description":"How to run IBM Granite-4.0 with Unsloth GGUFs on llama.cpp, Ollama and how to fine-tune!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"r2sc8WhniElY6ZuO8rZ9","title":"Llama 4: How to Run & Fine-tune","pathname":"/docs/models/tutorials/llama-4-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f999","description":"How to run Llama 4 locally using our dynamic GGUFs which recovers accuracy compared to standard quantization.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"efAr1xJJgeUlT0oMibEg","title":"Grok 2","pathname":"/docs/models/tutorials/grok-2","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"square-x-twitter","description":"Run xAI's Grok 2 model locally!","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"haJMAru7E9BKIjoJvoam","title":"Devstral: How to Run & Fine-tune","pathname":"/docs/models/tutorials/devstral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4d9","description":"Run and fine-tune Mistral Devstral 1.1, including Small-2507 and 2505.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"tYzsGxBMlN0JKeaQktNs","title":"How to Run Local LLMs with Docker: Step-by-Step Guide","pathname":"/docs/models/tutorials/how-to-run-llms-with-docker","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"docker","description":"Learn how to run Large Language Models (LLMs) with Docker & Unsloth on your local device.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"hGkhBczYQEpW4XkRzlks","title":"DeepSeek-V3-0324: How to Run Locally","pathname":"/docs/models/tutorials/deepseek-v3-0324-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f433","description":"How to run DeepSeek-V3-0324 locally using our dynamic quants which recovers accuracy","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"9a6TzBKnfHALYRRFfeNU","title":"DeepSeek-R1: How to Run Locally","pathname":"/docs/models/tutorials/deepseek-r1-how-to-run-locally","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f40b","description":"A guide on how you can run our 1.58-bit Dynamic Quants for DeepSeek-R1 using llama.cpp.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"NS98qKy46bSVzgzTOyzG","title":"DeepSeek-R1 Dynamic 1.58-bit","pathname":"/docs/models/tutorials/deepseek-r1-how-to-run-locally/deepseek-r1-dynamic-1.58-bit","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f433","description":"See performance comparison tables for Unsloth's Dynamic GGUF Quants vs Standard IMatrix Quants.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"},{"label":"DeepSeek-R1: How to Run Locally","emoji":"1f40b"}]},{"id":"zLTUUiWU5VchfMnYKD44","title":"Phi-4 Reasoning: How to Run & Fine-tune","pathname":"/docs/models/tutorials/phi-4-reasoning-how-to-run-and-fine-tune","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"windows","description":"Learn to run & fine-tune Phi-4 reasoning models locally with Unsloth + our Dynamic 2.0 quants","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"yWydsLIpPKmwwG1WDiKx","title":"QwQ-32B: How to Run effectively","pathname":"/docs/models/tutorials/qwq-32b-how-to-run-effectively","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f320","description":"How to run QwQ-32B effectively with our bug fixes and without endless generations + GGUFs.","breadcrumbs":[{"label":"Models"},{"label":"Large language model (LLMs) Tutorials","emoji":"1f680"}]},{"id":"7sCtc6YWnJBYthTjQsr7","title":"How to use Unsloth as an API endpoint","pathname":"/docs/basics/api","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"globe","description":"","breadcrumbs":[{"label":"Basics"}]},{"id":"gEugERiAw2ztDNt98JVR","title":"Inference & Deployment","pathname":"/docs/basics/inference-and-deployment","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f5a5","description":"Learn how to save your finetuned model so you can run it in your favorite inference engine.","breadcrumbs":[{"label":"Basics"}]},{"id":"T7ZPf3SNAwDykZNgXptE","title":"Saving to GGUF","pathname":"/docs/basics/inference-and-deployment/saving-to-gguf","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"}]},{"id":"KN9fwufcUfjR8cqPfQA4","title":"Speculative Decoding","pathname":"/docs/basics/inference-and-deployment/saving-to-gguf/speculative-decoding","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"Speculative Decoding with llama-server, llama.cpp, vLLM and more for 2x faster inference","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"},{"label":"Saving to GGUF"}]},{"id":"fhJtaLFFXVsGnbMUiACo","title":"vLLM Deployment & Inference Guide","pathname":"/docs/basics/inference-and-deployment/vllm-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"Guide on saving and deploying LLMs to vLLM for serving LLMs in production","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"}]},{"id":"T8vAb3VMIaDyIUlVtrdK","title":"vLLM Engine Arguments","pathname":"/docs/basics/inference-and-deployment/vllm-guide/vllm-engine-arguments","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"},{"label":"vLLM Deployment & Inference Guide"}]},{"id":"mp9Evu7eg8kdy0IfITHu","title":"LoRA Hot Swapping Guide","pathname":"/docs/basics/inference-and-deployment/vllm-guide/lora-hot-swapping-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"},{"label":"vLLM Deployment & Inference Guide"}]},{"id":"WehjgbuawqCXogREXvGG","title":"SGLang Deployment & Inference Guide","pathname":"/docs/basics/inference-and-deployment/sglang-guide","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"Guide on saving and deploying LLMs to SGLang for serving LLMs in production","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"}]},{"id":"wqYCVI9bC4YRS7jlwWhR","title":"llama-server & OpenAI endpoint Deployment Guide","pathname":"/docs/basics/inference-and-deployment/llama-server-and-openai-endpoint","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"Deploying via llama-server with an OpenAI compatible endpoint","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"}]},{"id":"KGLcoYbGfxvjuMoQ3HrB","title":"How to Run and Deploy LLMs on your iOS or Android Phone","pathname":"/docs/basics/inference-and-deployment/deploy-llms-phone","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4f1","description":"Tutorial for fine-tuning your own LLM and deploying it on your Android or iPhone with ExecuTorch.","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"}]},{"id":"e09iCJirEAJOrGDlyLre","title":"Troubleshooting Inference","pathname":"/docs/basics/inference-and-deployment/troubleshooting-inference","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"If you're experiencing issues when running or saving your model.","breadcrumbs":[{"label":"Basics"},{"label":"Inference & Deployment","emoji":"1f5a5"}]},{"id":"w020xJgdCTBtTvfHtvye","title":"How to Run Local LLMs with Claude Code","pathname":"/docs/basics/claude-code","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"claude","description":"Guide to use open models with Claude Code on your local device.","breadcrumbs":[{"label":"Basics"}]},{"id":"PCjZ57h5pE0QccKyJMYD","title":"How to Run Local LLMs with OpenAI Codex","pathname":"/docs/basics/codex","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"openai","description":"Use open models with OpenAI Codex on your device locally.","breadcrumbs":[{"label":"Basics"}]},{"id":"VTpEwCPKRuHGuVtz9ajd","title":"Multi-GPU Fine-tuning with Unsloth","pathname":"/docs/basics/multi-gpu-training-with-unsloth","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"rectangle-history","description":"Learn how to fine-tune LLMs on multiple GPUs and parallelism with Unsloth.","breadcrumbs":[{"label":"Basics"}]},{"id":"6aAeKwk7YLLgpBJR9JTv","title":"Multi-GPU Fine-tuning with Distributed Data Parallel (DDP)","pathname":"/docs/basics/multi-gpu-training-with-unsloth/ddp","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"Learn how to use the Unsloth CLI to train on multiple GPUs with Distributed Data Parallel (DDP)!","breadcrumbs":[{"label":"Basics"},{"label":"Multi-GPU Fine-tuning with Unsloth","icon":"rectangle-history"}]},{"id":"VlJjz852gxI14pEHjByu","title":"Fine-tuning Embedding Models with Unsloth Guide","pathname":"/docs/basics/embedding-finetuning","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f50e","description":"Learn how to easily fine-tune embedding models with Unsloth.","breadcrumbs":[{"label":"Basics"}]},{"id":"QLaEGK4QyFj4hjb4Cjdn","title":"Fine-tune MoE Models 12x Faster with Unsloth","pathname":"/docs/basics/faster-moe","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f48e","description":"Train MoE LLMs locally using Unsloth Guide.","breadcrumbs":[{"label":"Basics"}]},{"id":"bnULxAhnvp7EPaivWLiP","title":"Text-to-Speech (TTS) Fine-tuning Guide","pathname":"/docs/basics/text-to-speech-tts-fine-tuning","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f50a","description":"Learn how to to fine-tune TTS & STT voice models with Unsloth.","breadcrumbs":[{"label":"Basics"}]},{"id":"QznsvWxKKvrY6PdiByzz","title":"Unsloth Dynamic 2.0 GGUFs","pathname":"/docs/basics/unsloth-dynamic-2.0-ggufs","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9a5","description":"A big new upgrade to our Dynamic Quants!","breadcrumbs":[{"label":"Basics"}]},{"id":"jiaMFU5NqiW6tuZHwXPV","title":"Unsloth Dynamic GGUFs on Aider Polyglot","pathname":"/docs/basics/unsloth-dynamic-2.0-ggufs/unsloth-dynamic-ggufs-on-aider-polyglot","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f9a5","description":"Performance of Unsloth Dynamic GGUFs on Aider Polyglot Benchmarks","breadcrumbs":[{"label":"Basics"},{"label":"Unsloth Dynamic 2.0 GGUFs","emoji":"1f9a5"}]},{"id":"pEl4DasFmbHe97o5vK4R","title":"Tool Calling Guide for Local LLMs","pathname":"/docs/basics/tool-calling-guide-for-local-llms","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"screwdriver-wrench","description":"","breadcrumbs":[{"label":"Basics"}]},{"id":"Zo93wHRqGnzGRE62C395","title":"Vision Fine-tuning","pathname":"/docs/basics/vision-fine-tuning","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f441","description":"Learn how to fine-tune vision/multimodal LLMs with Unsloth","breadcrumbs":[{"label":"Basics"}]},{"id":"YCMcHSSIKR38pPhdq87W","title":"Troubleshooting & FAQs","pathname":"/docs/basics/troubleshooting-and-faqs","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"26a0","description":"Tips to solve issues, and frequently asked questions.","breadcrumbs":[{"label":"Basics"}]},{"id":"oJpfeo6SXe5svovA0sP6","title":"Hugging Face Hub, XET debugging","pathname":"/docs/basics/troubleshooting-and-faqs/hugging-face-hub-xet-debugging","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"Debugging, troubleshooting stalled, stuck downloads and slow downloads","breadcrumbs":[{"label":"Basics"},{"label":"Troubleshooting & FAQs","emoji":"26a0"}]},{"id":"kuRwnzkO5NvMH5METpT8","title":"Chat Templates","pathname":"/docs/basics/chat-templates","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f4ac","description":"Learn the fundamentals and customization options of chat templates, including Conversational, ChatML, ShareGPT, Alpaca formats, and more!","breadcrumbs":[{"label":"Basics"}]},{"id":"IklKRZMgbD798u0IJAQR","title":"Unsloth Environment Flags","pathname":"/docs/basics/unsloth-environment-flags","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f6e0","description":"Advanced flags which might be useful if you see breaking finetunes, or you want to turn stuff off.","breadcrumbs":[{"label":"Basics"}]},{"id":"8LEeKIVgEgdecg1B7Ahm","title":"Continued Pretraining","pathname":"/docs/basics/continued-pretraining","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"267b","description":"AKA as Continued Finetuning. Unsloth allows you to continually pretrain so a model can learn a new language.","breadcrumbs":[{"label":"Basics"}]},{"id":"IqlfM9k57UJVtsZ2xtRu","title":"Finetuning from Last Checkpoint","pathname":"/docs/basics/finetuning-from-last-checkpoint","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"1f3c1","description":"Checkpointing allows you to save your finetuning progress so you can pause it and then continue.","breadcrumbs":[{"label":"Basics"}]},{"id":"4iT5ojqRctQslDUikk5N","title":"Connect API Providers & Model Servers to Unsloth","pathname":"/docs/integrations/connections","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"cloud","description":"Guide to connect OpenAI, Anthropic, Ollama, llama.cpp, vLLM and other providers to Unsloth. Add API keys or model server URLs, load models, and use external models in chat.","breadcrumbs":[{"label":"Integrations"}]},{"id":"x0vCXW6AqnOAwV1wFylm","title":"Connect OpenAI to Unsloth: Run GPT Models in Local Chat","pathname":"/docs/integrations/connections/openai","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"","breadcrumbs":[{"label":"Integrations"},{"label":"Connect API Providers & Model Servers to Unsloth","icon":"cloud"}]},{"id":"O2Zs7Zzeudcv0mGLR74a","title":"Connect Anthropic to Unsloth: Run Claude Models in Local Chat","pathname":"/docs/integrations/connections/anthropic-claude","siteSpaceId":"sitesp_VHa4A","lang":"en","breadcrumbs":[{"label":"Integrations"},{"label":"Connect API Providers & Model Servers to Unsloth","icon":"cloud"}]},{"id":"QRKEvrTUSlnIqpY8RL2e","title":"Connect llama.cpp to Unsloth: Run GGUFs with llama-server","pathname":"/docs/integrations/connections/connect-llama.cpp-to-unsloth-run-ggufs-with-llama-server","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"","breadcrumbs":[{"label":"Integrations"},{"label":"Connect API Providers & Model Servers to Unsloth","icon":"cloud"}]},{"id":"TQ6YLngkMY6bBficXchP","title":"Connect vLLM to Unsloth for Local Chat Inference","pathname":"/docs/integrations/connections/vllm","siteSpaceId":"sitesp_VHa4A","lang":"en","breadcrumbs":[{"label":"Integrations"},{"label":"Connect API Providers & Model Servers to Unsloth","icon":"cloud"}]},{"id":"bRVbiO9tU82jB2cbrhOJ","title":"How to Connect Ollama to Unsloth","pathname":"/docs/integrations/connections/ollama","siteSpaceId":"sitesp_VHa4A","lang":"en","description":"","breadcrumbs":[{"label":"Integrations"},{"label":"Connect API Providers & Model Servers to Unsloth","icon":"cloud"}]},{"id":"BFxPZ1osvyZWS6VGk2UI","title":"How to Connect OpenRouter to Unsloth: API Key & Model Setup","pathname":"/docs/integrations/connections/openrouter","siteSpaceId":"sitesp_VHa4A","lang":"en","breadcrumbs":[{"label":"Integrations"},{"label":"Connect API Providers & Model Servers to Unsloth","icon":"cloud"}]},{"id":"CwQEpEmkKPmyEYdnEngt","title":"How to Run Local AI Models with OpenClaw","pathname":"/docs/integrations/openclaw","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"lobster","description":"Guide to running local LLMs with OpenClaw.","breadcrumbs":[{"label":"Integrations"}]},{"id":"qaA8ZjTxsH2GTuBOHyra","title":"How to Run Local AI Models with OpenCode","pathname":"/docs/integrations/openclaw/opencode","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"rectangle-vertical","description":"Guide to connect open LLMs with OpenCode on your local device.","breadcrumbs":[{"label":"Integrations"},{"label":"How to Run Local AI Models with OpenClaw","icon":"lobster"}]},{"id":"q1ZbCTKGY7P8eXLeDdEN","title":"How to Run Local AI Models with Hermes Agent","pathname":"/docs/integrations/hermes-agent","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"caduceus","description":"Guide on using open LLMs with Hermes Agent locally.","breadcrumbs":[{"label":"Integrations"}]},{"id":"viZvzp58ObzZkXtCm0qv","title":"Connect Python SDK to Unsloth","pathname":"/docs/integrations/connect-python-sdk-to-unsloth","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"python","description":"Guide to calling Unsloth's local API from Python using the official OpenAI or Anthropic SDKs including streaming, vision, function calling, and Unsloth's built-in server-side tools.","breadcrumbs":[{"label":"Integrations"}]},{"id":"1cwX0SOqPoqLx7fQ2sIS","title":"Connect Curl & HTTP to Unsloth","pathname":"/docs/integrations/connect-curl-and-http-to-unsloth","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"spiral","description":"Guide to hitting Unsloth's API with curl (or any HTTP client), complete with copy-pasteable recipes for every endpoint and feature..","breadcrumbs":[{"label":"Integrations"}]},{"id":"zb5JIG4b4Jhv2FaREe5I","title":"3x Faster LLM Training with Unsloth Kernels + Packing","pathname":"/docs/blog/3x-faster-training-packing","siteSpaceId":"sitesp_VHa4A","lang":"en","emoji":"26a1","description":"Learn how Unsloth increases training throughput and eliminates padding waste for fine-tuning.","breadcrumbs":[{"label":"Blog"}]},{"id":"bNA4JFPPNIXhfmQY7JfQ","title":"500K Context Length Fine-tuning","pathname":"/docs/blog/500k-context-length-fine-tuning","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"ruler-combined","description":"Learn how to enable >500K token context window fine-tuning with Unsloth.","breadcrumbs":[{"label":"Blog"}]},{"id":"OukZvlYQT7UIMT2ULKav","title":"Quantization-Aware Training (QAT)","pathname":"/docs/blog/quantization-aware-training-qat","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"down-left-and-up-right-to-center","description":"Quantize models to 4-bit with Unsloth and PyTorch to recover accuracy.","breadcrumbs":[{"label":"Blog"}]},{"id":"g1UMvrOXMWVheqG9egap","title":"Fine-Tuning LLMs on NVIDIA DGX Station with Unsloth","pathname":"/docs/blog/dgx-station","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"microchip-ai","description":"NVIDIA DGX Station tutorial on how to fine-tune with notebooks from Unsloth.","breadcrumbs":[{"label":"Blog"}]},{"id":"cP88TERn3hrjTC46YULf","title":"How to Fine-tune LLMs with Unsloth & Docker","pathname":"/docs/blog/how-to-fine-tune-llms-with-unsloth-and-docker","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"docker","description":"Learn how to fine-tune LLMs or do Reinforcement Learning (RL) with Unsloth's Docker image.","breadcrumbs":[{"label":"Blog"}]},{"id":"SP7tOaHurV8iKXitTy2O","title":"Fine-tuning LLMs with NVIDIA DGX Spark and Unsloth","pathname":"/docs/blog/fine-tuning-llms-with-nvidia-dgx-spark-and-unsloth","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"sparkle","description":"Tutorial on how to fine-tune and do reinforcement learning (RL) with OpenAI gpt-oss on NVIDIA DGX Spark.","breadcrumbs":[{"label":"Blog"}]},{"id":"FmZGHW74OKsd4nalCidC","title":"Fine-tuning LLMs with Blackwell, RTX 50 series & Unsloth","pathname":"/docs/blog/fine-tuning-llms-with-blackwell-rtx-50-series-and-unsloth","siteSpaceId":"sitesp_VHa4A","lang":"en","icon":"microchip","description":"Learn how to fine-tune LLMs on NVIDIA's Blackwell RTX 50 series and B200 GPUs with our step-by-step guide.","breadcrumbs":[{"label":"Blog"}]},{"id":"d223e1d473d91b0a823446588fad0295a03dee2e","title":"Unsloth 文档","pathname":"/docs/zh","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9a5","description":"Unsloth 是一个用于运行和训练模型的开源框架。","breadcrumbs":[{"label":"开始使用"}]},{"id":"237026badf5bc822cff5a01118fb7de3da2e2153","title":"面向初学者的微调","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-for-beginners","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"2b50","description":"","breadcrumbs":[{"label":"开始使用"}]},{"id":"fad5753068629237f814fd7a2737bf5277187b58","title":"Unsloth 要求","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-for-beginners/unsloth-requirements","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f6e0","description":"这里列出了 Unsloth 的要求，包括系统和 GPU VRAM 要求。","breadcrumbs":[{"label":"开始使用"},{"label":"面向初学者的微调","emoji":"2b50"}]},{"id":"23f746d97a493793f0cf778956c3d1e708c15940","title":"常见问题 + 微调适合我吗？","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-for-beginners/faq-+-is-fine-tuning-right-for-me","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f914","description":"如果你还在犹豫微调是否适合你，看看这里！了解微调的误区，以及它与 RAG 的比较等内容：","breadcrumbs":[{"label":"开始使用"},{"label":"面向初学者的微调","emoji":"2b50"}]},{"id":"622af8e56d0924dd99cc871bc54a538197574e1a","title":"Unsloth 笔记本","pathname":"/docs/zh/kai-shi-shi-yong/unsloth-notebooks","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4d2","description":"微调笔记本：浏览 Unsloth 目录。","breadcrumbs":[{"label":"开始使用"}]},{"id":"20805f6881460e7c3a088cff24acf0f1090f1984","title":"Unsloth 模型目录","pathname":"/docs/zh/kai-shi-shi-yong/unsloth-model-catalog","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f52e","description":"","breadcrumbs":[{"label":"开始使用"}]},{"id":"32a7ce415cf3da2e50fd777381735610343fea45","title":"Unsloth 安装","pathname":"/docs/zh/kai-shi-shi-yong/install","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4e5","description":"了解如何在本地或在线安装 Unsloth。","breadcrumbs":[{"label":"开始使用"}]},{"id":"4597301c2f6c6df2660ec569f5a1d9e8481eb494","title":"通过 pip 和 uv 安装 Unsloth","pathname":"/docs/zh/kai-shi-shi-yong/install/pip-install","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"desktop-arrow-down","description":"要通过 Pip 在本地安装 Unsloth，请按照以下步骤操作：","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"9e9449bc388de91ddda522975936cef81b928bce","title":"在 MacOS 上安装 Unsloth","pathname":"/docs/zh/kai-shi-shi-yong/install/mac","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"apple","description":"","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"bc3c53c3e0f87180567dda04ad25f8d2b13374ce","title":"如何在 Windows 上使用 Unsloth 对 LLM 进行微调（分步指南）","pathname":"/docs/zh/kai-shi-shi-yong/install/windows-installation","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"windows","description":"了解如何在 Windows 上安装 Unsloth，以开始在本地对 LLM 进行微调。","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"2143ac5379e70861b75a6ff27ff09a4a4c8034c5","title":"通过 Docker 安装 Unsloth","pathname":"/docs/zh/kai-shi-shi-yong/install/docker","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"docker","description":"使用我们的官方 Docker 容器安装 Unsloth","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"754d57d7c027164b0f10301137e5ddd66e802e42","title":"更新 Unsloth","pathname":"/docs/zh/kai-shi-shi-yong/install/updating","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"arrow-rotate-right","description":"要更新或使用旧版本的 Unsloth，请按照以下步骤操作：","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"58db76de3e4da663680e51bb3a7f48ee9575ee07","title":"使用 Unsloth 在 AMD GPU 上微调 LLM 指南","pathname":"/docs/zh/kai-shi-shi-yong/install/amd","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"square-up-right","description":"了解如何使用 Unsloth 在 AMD GPU 上微调大型语言模型（LLM）。","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"fb0bfd00002552fa929669fb13c5aafab2f274d1","title":"使用 Unsloth 在 Intel GPU 上微调 LLM","pathname":"/docs/zh/kai-shi-shi-yong/install/intel","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"info","description":"了解如何在 Intel GPU 上训练和微调大型语言模型。","breadcrumbs":[{"label":"开始使用"},{"label":"Unsloth 安装","emoji":"1f4e5"}]},{"id":"e722e86e330786e5915445b91f900d9f9e0ba067","title":"LLM 微调指南","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-llms-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9ec","description":"学习微调的所有基础知识和最佳实践，适合初学者。","breadcrumbs":[{"label":"开始使用"}]},{"id":"e3d8f41874867d09291eadcea96a34bf92b096a7","title":"数据集指南","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-llms-guide/datasets-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4c8","description":"学习如何创建和准备用于微调的数据集。","breadcrumbs":[{"label":"开始使用"},{"label":"LLM 微调指南","emoji":"1f9ec"}]},{"id":"0c67f9d3a44c115b4fda319de5774cc3513ea3a3","title":"LoRA 微调超参数指南","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-llms-guide/lora-hyperparameters-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9e0","description":"逐步了解最佳的 LLM 微调设置——LoRA rank 和 alpha、训练轮数、批量大小 + 梯度累积、QLoRA 与 LoRA、目标模块等。","breadcrumbs":[{"label":"开始使用"},{"label":"LLM 微调指南","emoji":"1f9ec"}]},{"id":"fb27a5a49eec8190a033600d7b4f8f78f333f7a9","title":"我应该使用什么模型进行微调？","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-llms-guide/what-model-should-i-use","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"2753","description":"","breadcrumbs":[{"label":"开始使用"},{"label":"LLM 微调指南","emoji":"1f9ec"}]},{"id":"e36b62f99558f2c1e1e644ea9bc15aa9621d449b","title":"教程：如何微调 Llama-3 并在 Ollama 中使用","pathname":"/docs/zh/kai-shi-shi-yong/fine-tuning-llms-guide/tutorial-how-to-finetune-llama-3-and-use-in-ollama","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f999","description":"面向初学者的指南：创建一个可在 Ollama 上本地运行的个性化私人助手（类似 ChatGPT）","breadcrumbs":[{"label":"开始使用"},{"label":"LLM 微调指南","emoji":"1f9ec"}]},{"id":"7e8be22a966d3861ff1f7ebbd178ae7144d05c51","title":"强化学习（RL）指南","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4a1","description":"全面了解强化学习（RL），以及如何使用 Unsloth 和 GRPO 训练你自己的 DeepSeek-R1 推理模型。从入门到进阶的完整指南。","breadcrumbs":[{"label":"开始使用"}]},{"id":"568f528390785432e96ab3958c36bec4b8482ec7","title":"使用 7 倍更长上下文的强化学习 GRPO","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/grpo-long-context","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f300","description":"了解 Unsloth 如何实现超长上下文的 RL 微调。","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"1d7550c0346dc79003c120cca2e0063a9addf768","title":"视觉强化学习（VLM RL）","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/vision-reinforcement-learning-vlm-rl","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f441-1f5e8","description":"通过 GRPO 和 RL 使用 Unsloth 训练视觉/多模态模型！","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"f63218900ca62905f408a7e5cca35eb4279231d7","title":"FP8 强化学习","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/fp8-reinforcement-learning","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f3b1","description":"使用 Unsloth 以 FP8 精度训练强化学习（RL）和 GRPO。","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"289d4338990d56fc4202e77bbae62bb949a7b439","title":"教程：使用 GRPO 训练你自己的推理模型","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/tutorial-train-your-own-reasoning-model-with-grpo","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"26a1","description":"面向初学者的指南：通过使用 Unsloth 和 GRPO，将类似 Llama 3.1（8B）的模型转变为推理模型。","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"6223874117c54cbdacd6f3467136ab1146fce98a","title":"高级强化学习文档","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/advanced-rl-documentation","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9e9","description":"在使用 Unsloth 和 GRPO 时的高级文档设置。","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"20f0796d08c62347da8d70300eda483d376f94bb","title":"GSPO 强化学习","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/advanced-rl-documentation/gspo-reinforcement-learning","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"lightbulb-on","description":"在 Unsloth 中使用 GSPO（组序列策略优化）进行 RL 训练。","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"},{"label":"高级强化学习文档","emoji":"1f9e9"}]},{"id":"5aa66df0eeb87307aba670f22bb02189b4d9e37f","title":"RL 奖励黑客","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/advanced-rl-documentation/rl-reward-hacking","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"treasure-chest","description":"了解什么是强化学习中的奖励黑客，以及如何应对它。","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"},{"label":"高级强化学习文档","emoji":"1f9e9"}]},{"id":"5487fa62e1aa76e8bf480388ee8d3305a4f0c4e1","title":"RL 中的 FP16 与 BF16","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/advanced-rl-documentation/fp16-vs-bf16-for-rl","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"2049","description":"通过 FP16 击败训练-推理不匹配 https://arxiv.org/pdf/2510.26788 表明使用 float16 比 bfloat16 更好","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"},{"label":"高级强化学习文档","emoji":"1f9e9"}]},{"id":"838ad11f4c8ab2c80a5f0297ee29a596bf49a0e6","title":"内存高效型 RL","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/memory-efficient-rl","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"memory","description":"","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"de455e62732ad7811d23f32a140e457c7d66724d","title":"偏好优化训练 - DPO、ORPO 和 KTO","pathname":"/docs/zh/kai-shi-shi-yong/reinforcement-learning-rl-guide/preference-dpo-orpo-and-kto","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f3c6","description":"通过 Unsloth 了解使用 DPO、GRPO、ORPO 或 KTO 进行偏好对齐微调，请按照以下步骤操作：","breadcrumbs":[{"label":"开始使用"},{"label":"强化学习（RL）指南","emoji":"1f4a1"}]},{"id":"1bc06634f9646051a40b3ee9c1e88ef2179a0bc2","title":"Unsloth Studio 介绍","pathname":"/docs/zh/xin/studio","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9a5","description":"使用 Unsloth Studio 在本地运行和训练 AI 模型。","breadcrumbs":[{"label":"新"}]},{"id":"12f6b822ad28322267b74accad74f665b341f5a4","title":"Unsloth Studio 入门","pathname":"/docs/zh/xin/studio/start","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"bolt","description":"一份关于微调工作室、数据配方、模型导出和聊天功能的入门指南。","breadcrumbs":[{"label":"新"},{"label":"Unsloth Studio 介绍","emoji":"1f9a5"}]},{"id":"5c2325084ff65c0303d8ec102b689868935855d3","title":"如何使用 Unsloth Studio 运行模型","pathname":"/docs/zh/xin/studio/chat","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"comment-dots","description":"使用 Unsloth Studio 在本地运行 AI 模型、LLM 和 GGUF。","breadcrumbs":[{"label":"新"},{"label":"Unsloth Studio 介绍","emoji":"1f9a5"}]},{"id":"ef666f465ed5ec55124d17ba90cfb169b4efa95f","title":"Unsloth Studio 安装","pathname":"/docs/zh/xin/studio/install","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"arrow-down-to-square","description":"了解如何在本地设备上安装 Unsloth Studio。","breadcrumbs":[{"label":"新"},{"label":"Unsloth Studio 介绍","emoji":"1f9a5"}]},{"id":"7223afcfd2df87e1fe32a963c0b5ef0e45f563c5","title":"Unsloth 数据配方","pathname":"/docs/zh/xin/studio/data-recipe","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"hat-chef","description":"了解如何使用 Unsloth Studio 的数据配方创建、构建和编辑数据集。","breadcrumbs":[{"label":"新"},{"label":"Unsloth Studio 介绍","emoji":"1f9a5"}]},{"id":"f7c3389bdba9af3050e66a941596d827cdb11e0b","title":"使用 Unsloth Studio 导出模型","pathname":"/docs/zh/xin/studio/export","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"box-isometric","description":"了解如何将你的 safetensor 或 LoRA 模型文件导出为 GGUF 或其他格式。","breadcrumbs":[{"label":"新"},{"label":"Unsloth Studio 介绍","emoji":"1f9a5"}]},{"id":"e1e43893beb1c3e2a075324e9a00800315b2e1a3","title":"Unsloth 更新","pathname":"/docs/zh/xin/changelog","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"sparkles","description":"Unsloth 最新版本、改进和修复的更新日志。","breadcrumbs":[{"label":"新"}]},{"id":"213bd08e4302b621f4392f7ee38decb275ffab02","title":"Qwen3.6 - 如何本地运行","pathname":"/docs/zh/mo-xing/qwen3.6","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f49c","description":"在本地运行新的 Qwen3.6-27B 和 35B-A3B 模型！","breadcrumbs":[{"label":"模型"}]},{"id":"10f714f4a513e0d0a86b6f9d5945f9014729b035","title":"Gemma 4 - 如何本地运行","pathname":"/docs/zh/mo-xing/gemma-4","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"2728","description":"在本地运行 Google 的新 Gemma 4 模型，包括 E2B、E4B、26B A4B 和 31B。","breadcrumbs":[{"label":"模型"}]},{"id":"33fa9e3bb3ccf6a5c0011aa600e98abbe3a829e3","title":"Gemma 4 微调指南","pathname":"/docs/zh/mo-xing/gemma-4/train","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"flask-gear","description":"使用 Unsloth 训练 Google 的 Gemma 4。","breadcrumbs":[{"label":"模型"},{"label":"Gemma 4 - 如何本地运行","emoji":"2728"}]},{"id":"31f07ee6a66a73019ccc7bc333592f8522540f7b","title":"NVIDIA Nemotron 3 Nano Omni - 如何本地运行","pathname":"/docs/zh/mo-xing/nemotron-3-nano-omni","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9e9","description":"在你的设备上本地运行并微调 Nemotron-3-Nano-Omni-30B-A3B！","breadcrumbs":[{"label":"模型"}]},{"id":"8a1261a3b4582d901fbe0ea49e315e28824eebf2","title":"Kimi K2.6 - 如何本地运行","pathname":"/docs/zh/mo-xing/kimi-k2.6","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f95d","description":"在你自己的本地设备上运行 Kimi-K2.6 的分步指南。","breadcrumbs":[{"label":"模型"}]},{"id":"1427becb679b955148197a06de42a82ae44b05b6","title":"Qwen3.5 - 如何本地运行","pathname":"/docs/zh/mo-xing/qwen3.5","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f49c","description":"在你的本地设备上运行新的 Qwen3.5 LLM，包括 Medium：Qwen3.5-35B-A3B、27B、122B-A10B，Small：Qwen3.5-0.8B、2B、4B、9B 和 397B-A17B！","breadcrumbs":[{"label":"模型"}]},{"id":"f5ae6c6f6b6b3c34b616c7d668668d9ab102aa23","title":"Qwen3.5 微调指南","pathname":"/docs/zh/mo-xing/qwen3.5/fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"flask-gear","description":"了解如何使用 Unsloth 微调 Qwen3.5 LLM。","breadcrumbs":[{"label":"模型"},{"label":"Qwen3.5 - 如何本地运行","emoji":"1f49c"}]},{"id":"29170937075312be229b292fa371d86315687849","title":"Qwen3.5 GGUF 基准测试","pathname":"/docs/zh/mo-xing/qwen3.5/gguf-benchmarks","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"chart-fft","description":"查看 Unsloth Dynamic GGUF 的表现，以及对困惑度、KL 散度和 MXFP4 的分析。","breadcrumbs":[{"label":"模型"},{"label":"Qwen3.5 - 如何本地运行","emoji":"1f49c"}]},{"id":"9aaee5fe955235c67f90c9c0fd454b487c5f2d80","title":"GLM-5.1 - 如何本地运行","pathname":"/docs/zh/mo-xing/glm-5.1","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"z","description":"在你自己的本地设备上运行 Z.ai 的新 GLM-5.1 模型！","breadcrumbs":[{"label":"模型"}]},{"id":"8d4117b244a368b8f80b1a9d079fa31360c8e823","title":"大型语言模型（LLM）教程","pathname":"/docs/zh/mo-xing/tutorials","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f680","description":"","breadcrumbs":[{"label":"模型"}]},{"id":"ea62c0023475466d68867497372502852aec44f9","title":"Qwen3 - 如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/qwen3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f320","description":"学习如何使用 Unsloth 和我们的 Dynamic 2.0 量化在本地运行和微调 Qwen3","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"27804e7e4efe89993216789d2932c5b1415cf077","title":"Qwen3-VL：如何运行指南","pathname":"/docs/zh/mo-xing/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-vl-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f320","description":"学习如何使用 Unsloth 在本地微调和运行 Qwen3-VL。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"},{"label":"Qwen3 - 如何运行和微调","emoji":"1f320"}]},{"id":"7b9f9f1f51740c3843cc4f6e4c00590de39b488b","title":"Qwen3-2507：本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-2507","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f320","description":"在你的设备上本地运行 Qwen3-30B-A3B-2507 和 235B-A22B 的 Thinking 与 Instruct 版本！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"},{"label":"Qwen3 - 如何运行和微调","emoji":"1f320"}]},{"id":"9bee4aaebd8150d5123bcdff6d78c7cc9a85098b","title":"MiniMax-M2.7 - 如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/minimax-m27","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"waveform","description":"在你自己的设备上本地运行 MiniMax-M2.7 LLM！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"b0620b25ba70d3f855acc525e8dbd8dc4c0e7acb","title":"GLM-5：如何本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/glm-5","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"z","description":"在你自己的本地设备上运行 Z.ai 的新 GLM-5 模型！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"5c26f8be53c96fd6b94596a72ef056d730c03d86","title":"Kimi K2.5：如何本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/kimi-k2.5","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f95d","description":"在你自己的本地设备上运行 Kimi-K2.5 的指南！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"a63fe18206ba278e17e40bf31ce577cda524869c","title":"GLM-4.7-Flash：如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/glm-4.7-flash","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"z","description":"在你的设备上本地运行并微调 GLM-4.7-Flash！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"77ac2f6d69532a9ed1df88fcd4048f113fedb4ad","title":"MiniMax-M2.5：如何运行指南","pathname":"/docs/zh/mo-xing/tutorials/minimax-m25","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"waveform","description":"在你自己的设备上本地运行 MiniMax-M2.5！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"04b53b5f8ccc52e6530ba353ac2df848a070f75e","title":"Qwen3-Coder：如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/qwen3-coder-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f320","description":"使用 Unsloth Dynamic 量化在本地运行 Qwen3-Coder-30B-A3B-Instruct 和 480B-A35B。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"d4856201ff076e83de868b083ab0653288b531ee","title":"Gemma 3 - 如何运行指南","pathname":"/docs/zh/mo-xing/tutorials/gemma-3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"google","description":"如何通过我们的 GGUF 在 llama.cpp、Ollama、Open WebUI 上高效运行 Gemma 3，以及如何使用 Unsloth 进行微调！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"4f67226331354d381a9545a411f8f9364fd9cf27","title":"Gemma 3n：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/gemma-3-how-to-run-and-fine-tune/gemma-3n-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"google","description":"在 llama.cpp、Ollama、Open WebUI 上使用 Dynamic GGUF 在本地运行 Google 的新 Gemma 3n，并用 Unsloth 进行微调！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"},{"label":"Gemma 3 - 如何运行指南","icon":"google"}]},{"id":"0efbaa992ff738764da5513b9bb33e8536c93397","title":"DeepSeek-OCR 2：如何运行和微调指南","pathname":"/docs/zh/mo-xing/tutorials/deepseek-ocr-2","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f433","description":"关于如何在本地运行和微调 DeepSeek-OCR-2 的指南。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"f2f10d2f6d415fee991d199028e61af190b1bd9f","title":"GLM-4.7：如何本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/glm-4.7","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"z","description":"关于如何在你自己的本地设备上运行 Z.ai GLM-4.7 模型的指南！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"b5c7d5b7677814317721b04088046f74558e7756","title":"如何在 ComfyUI 中本地运行 Qwen-Image-2512","pathname":"/docs/zh/mo-xing/tutorials/qwen-image-2512","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f49f","description":"在你的本地设备上使用 ComfyUI 运行 Qwen-Image-2512 的分步教程。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"36c45bfa260807b70cd911eddc881f78ab3166b4","title":"在 stable-diffusion.cpp 中运行 Qwen-Image-2512 教程","pathname":"/docs/zh/mo-xing/tutorials/qwen-image-2512/stable-diffusion.cpp","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f3a8","description":"在 stable-diffusion.cpp 中使用 Qwen-Image-2512 的教程。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"},{"label":"如何在 ComfyUI 中本地运行 Qwen-Image-2512","emoji":"1f49f"}]},{"id":"1e5e3967b00f3aee6ffa79790e93cf515cd7a47d","title":"Devstral 2 - 如何运行指南","pathname":"/docs/zh/mo-xing/tutorials/devstral-2","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4d9","description":"用于在本地运行 Mistral Devstral 2 模型的指南：123B-Instruct-2512 和 Small-2-24B-Instruct-2512。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"4991295da147774f87b981fb5f8227e572f31c2f","title":"Ministral 3 - 如何运行指南","pathname":"/docs/zh/mo-xing/tutorials/ministral-3","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f431","description":"Mistral Ministral 3 模型指南，可在你的设备上本地运行或微调","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"18f54403c2ec31fa43d1878d7b0d9f667b94a739","title":"DeepSeek-OCR：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/deepseek-ocr-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f433","description":"关于如何在本地运行和微调 DeepSeek-OCR 的指南。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"46d34a983e9c5a2763bb18c48daf83bcbd1ba871","title":"Kimi K2 Thinking：本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/kimi-k2-thinking-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f319","description":"在你自己的本地设备上运行 Kimi-K2-Thinking 和 Kimi-K2 的指南！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"c95883ea2c2bc5222ff12a80cc8e663fcb28cbdd","title":"GLM-4.6：本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/glm-4.6-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"z","description":"关于如何在你自己的本地设备上运行 Z.ai GLM-4.6 和 GLM-4.6V-Flash 模型的指南！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"8db617b6a4600d935dc1abb0de60011e111d55ff","title":"Qwen3-Next：本地运行指南","pathname":"/docs/zh/mo-xing/tutorials/qwen3-next","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f320","description":"在你的设备上本地运行 Qwen3-Next-80B-A3B-Instruct 和 Thinking 版本！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"585492ce9031c7bf11018debd8017628aaf0db18","title":"FunctionGemma：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/functiongemma","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"google","description":"学习如何在你的设备和手机上本地运行和微调 FunctionGemma。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"e159c2414bf59eef4f0d1e98dc6013b013c34532","title":"DeepSeek-V3.1：如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/deepseek-v3.1-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f40b","description":"关于如何在你自己的本地设备上运行 DeepSeek-V3.1 和 Terminus 的指南！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"cc90024db8d8f2b8456ab21f1a604bf6329b4ade","title":"DeepSeek-R1-0528：如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/deepseek-r1-0528-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f40b","description":"关于如何在你自己的本地设备上运行 DeepSeek-R1-0528，包括 Qwen3 的指南！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"ca88dbf750085097663e25264a22a8807b3765ef","title":"Liquid LFM2.5：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/lfm2.5","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4a7","description":"在你的设备上本地运行并微调 LFM2.5 Instruct 和 Vision！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"3e19eb045e2102b53e82b58d156a4c97ae6e6841","title":"Magistral：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/magistral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4a5","description":"认识 Magistral——Mistral 的新推理模型。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"68af53cc82ad32be2fd8d907ed61998fd4ddb80c","title":"IBM Granite 4.0","pathname":"/docs/zh/mo-xing/tutorials/ibm-granite-4.0","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"cube","description":"如何在 llama.cpp、Ollama 上使用 Unsloth GGUF 运行 IBM Granite-4.0，以及如何微调！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"42678e5a7e3e685fdb02fdfa055add8deb255047","title":"Llama 4：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/llama-4-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f999","description":"如何使用我们的动态 GGUF 在本地运行 Llama 4，与标准量化相比可恢复准确率。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"1834e3a31973190a1f05682cdcd1a9031464c344","title":"Grok 2","pathname":"/docs/zh/mo-xing/tutorials/grok-2","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"square-x-twitter","description":"在本地运行 xAI 的 Grok 2 模型！","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"bf18dbdaeec1eb9a55d550c53a565fcd383bbedc","title":"Devstral：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/devstral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4d9","description":"运行并微调 Mistral Devstral 1.1，包括 Small-2507 和 2505。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"cee0a9958d4af5692dc2cea39034597e9f71ddfd","title":"如何通过 Docker 运行本地 LLM：分步指南","pathname":"/docs/zh/mo-xing/tutorials/how-to-run-llms-with-docker","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"docker","description":"了解如何在你的本地设备上使用 Docker 和 Unsloth 运行大型语言模型（LLM）。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"475f6ff513257bac96c996ccb621236825757b4e","title":"DeepSeek-V3-0324：如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/deepseek-v3-0324-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f433","description":"如何使用我们的动态量化在本地运行 DeepSeek-V3-0324，以恢复准确率","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"9aeceb5c9d6f0a75d39bba30684b3f56d8aa245f","title":"DeepSeek-R1：如何本地运行","pathname":"/docs/zh/mo-xing/tutorials/deepseek-r1-how-to-run-locally","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f40b","description":"关于如何使用 llama.cpp 运行我们的 1.58-bit DeepSeek-R1 动态量化的指南。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"6a1a38ace50da481fbc7af2ec6d45ba6f2779db6","title":"DeepSeek-R1 动态 1.58-bit","pathname":"/docs/zh/mo-xing/tutorials/deepseek-r1-how-to-run-locally/deepseek-r1-dynamic-1.58-bit","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f433","description":"查看 Unsloth Dynamic GGUF 量化与标准 IMatrix 量化的性能对比表。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"},{"label":"DeepSeek-R1：如何本地运行","emoji":"1f40b"}]},{"id":"4af548c36209e6a98ab992e40cc8085cf71fe1b1","title":"Phi-4 推理：如何运行和微调","pathname":"/docs/zh/mo-xing/tutorials/phi-4-reasoning-how-to-run-and-fine-tune","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"windows","description":"学习如何使用 Unsloth 和我们的 Dynamic 2.0 量化在本地运行和微调 Phi-4 推理模型","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"152e63aa9c7c5dfb094c254c117d4bd06908bf98","title":"QwQ-32B：如何高效运行","pathname":"/docs/zh/mo-xing/tutorials/qwq-32b-how-to-run-effectively","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f320","description":"如何通过我们的修复补丁以及 GGUF 来高效运行 QwQ-32B，避免无限生成。","breadcrumbs":[{"label":"模型"},{"label":"大型语言模型（LLM）教程","emoji":"1f680"}]},{"id":"2c2bb53a273009e389791ded9e28dd4769a55051","title":"如何将 Unsloth 作为 API 端点使用","pathname":"/docs/zh/ji-chu/api","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"globe","description":"","breadcrumbs":[{"label":"基础"}]},{"id":"9a72670992feb75def412a693565c84a88c8a266","title":"推理与部署","pathname":"/docs/zh/ji-chu/inference-and-deployment","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f5a5","description":"了解如何保存你微调后的模型，以便在你喜欢的推理引擎中运行它。","breadcrumbs":[{"label":"基础"}]},{"id":"b83d88f106d75c3396c46f5342fb401501910093","title":"保存为 GGUF","pathname":"/docs/zh/ji-chu/inference-and-deployment/saving-to-gguf","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"8ff566c5ada3f8fb59929a32b877837ba0041924","title":"推测解码","pathname":"/docs/zh/ji-chu/inference-and-deployment/saving-to-gguf/speculative-decoding","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"使用 llama-server、llama.cpp、vLLM 等进行推测解码，实现 2 倍更快的推理","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"},{"label":"保存为 GGUF"}]},{"id":"9f0e22d200c9105481e4854b8473aba99ca44835","title":"vLLM 部署与推理指南","pathname":"/docs/zh/ji-chu/inference-and-deployment/vllm-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"关于将 LLM 保存并部署到 vLLM，以便在生产环境中提供 LLM 服务的指南","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"160443d79a06d2d700045d140452e790dbdb1173","title":"vLLM 引擎参数","pathname":"/docs/zh/ji-chu/inference-and-deployment/vllm-guide/vllm-engine-arguments","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"},{"label":"vLLM 部署与推理指南"}]},{"id":"5536133cb2c4c06df946ff9440b26b7391a12b5c","title":"LoRA 热插拔指南","pathname":"/docs/zh/ji-chu/inference-and-deployment/vllm-guide/lora-hot-swapping-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"},{"label":"vLLM 部署与推理指南"}]},{"id":"d4f9cf59ddb6cd217d8f8563eeb6c00042f21972","title":"保存到 Ollama","pathname":"/docs/zh/ji-chu/inference-and-deployment/saving-to-ollama","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"771775d41e6a1596232819cdd79823d415eda744","title":"将模型部署到 LM Studio","pathname":"/docs/zh/ji-chu/inference-and-deployment/lm-studio","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"将模型保存为 GGUF，以便你可以将其运行并部署到 LM Studio","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"b229232347528d9aad88114911fe039c42595fd3","title":"如何在 Linux 终端中安装 LM Studio CLI","pathname":"/docs/zh/ji-chu/inference-and-deployment/lm-studio/how-to-install-lm-studio-cli-in-linux-terminal","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f47e","description":"在终端实例中无需 UI 的 LM Studio CLI 安装指南。","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"},{"label":"将模型部署到 LM Studio"}]},{"id":"23e76b12a72496ba4fcc9d857dd940dd6ae14736","title":"SGLang 部署与推理指南","pathname":"/docs/zh/ji-chu/inference-and-deployment/sglang-guide","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"关于将 LLM 保存并部署到 SGLang，以便在生产环境中提供 LLM 服务的指南","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"4a201e2f3e992b62e25a0ba283ec8b14ad3f414b","title":"llama-server 与 OpenAI 端点部署指南","pathname":"/docs/zh/ji-chu/inference-and-deployment/llama-server-and-openai-endpoint","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"通过 llama-server 部署并提供兼容 OpenAI 的端点","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"e0e826e45659eab088ef3acd7826998bc36539e9","title":"如何在你的 iOS 或 Android 手机上运行和部署 LLM","pathname":"/docs/zh/ji-chu/inference-and-deployment/deploy-llms-phone","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4f1","description":"关于微调你自己的 LLM，并使用 ExecuTorch 将其部署到 Android 或 iPhone 的教程。","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"5511a85b8b57f4cfdffedbc8f1ea2110a10d550e","title":"推理故障排查","pathname":"/docs/zh/ji-chu/inference-and-deployment/troubleshooting-inference","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"如果你在运行或保存模型时遇到问题。","breadcrumbs":[{"label":"基础"},{"label":"推理与部署","emoji":"1f5a5"}]},{"id":"1a707991086189a8e5cd8374f3ce1b81915bc159","title":"如何使用 Claude Code 运行本地 LLM","pathname":"/docs/zh/ji-chu/claude-code","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"claude","description":"在你的本地设备上使用 Claude Code 运行开源模型的指南。","breadcrumbs":[{"label":"基础"}]},{"id":"b71ddea7924324c058a771e5e831c3cb6fc75b18","title":"如何使用 OpenAI Codex 运行本地 LLM","pathname":"/docs/zh/ji-chu/codex","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"openai","description":"在你的设备上本地使用 OpenAI Codex 运行开源模型。","breadcrumbs":[{"label":"基础"}]},{"id":"cd472d76f8ad81236011a0337b4cec66382031e2","title":"使用 Unsloth 进行多 GPU 微调","pathname":"/docs/zh/ji-chu/multi-gpu-training-with-unsloth","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"rectangle-history","description":"了解如何使用 Unsloth 在多 GPU 和并行环境下对 LLM 进行微调。","breadcrumbs":[{"label":"基础"}]},{"id":"c16d3ee721c7692f2341fe3dcb16afaf2d84858f","title":"使用分布式数据并行（DDP）进行多 GPU 微调","pathname":"/docs/zh/ji-chu/multi-gpu-training-with-unsloth/ddp","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"了解如何使用 Unsloth CLI 通过分布式数据并行（DDP）在多个 GPU 上训练！","breadcrumbs":[{"label":"基础"},{"label":"使用 Unsloth 进行多 GPU 微调","icon":"rectangle-history"}]},{"id":"a078f1a9ba6457ae124f908cfdebf7ca27afaf56","title":"使用 Unsloth 微调嵌入模型指南","pathname":"/docs/zh/ji-chu/embedding-finetuning","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f50e","description":"了解如何使用 Unsloth 轻松微调嵌入模型。","breadcrumbs":[{"label":"基础"}]},{"id":"90adfb72dc954d9c5afd7cd406bebf264f2005ac","title":"使用 Unsloth 将 MoE 模型微调速度提升 12 倍","pathname":"/docs/zh/ji-chu/faster-moe","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f48e","description":"使用 Unsloth 指南在本地训练 MoE LLM。","breadcrumbs":[{"label":"基础"}]},{"id":"d2ec5af816022e7e65a1929ffa5d6060bf270047","title":"文本转语音（TTS）微调指南","pathname":"/docs/zh/ji-chu/text-to-speech-tts-fine-tuning","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f50a","description":"了解如何使用 Unsloth 微调 TTS 和 STT 语音模型。","breadcrumbs":[{"label":"基础"}]},{"id":"e658f01212ed739b6cc1648a22333767661730a1","title":"Unsloth Dynamic 2.0 GGUF","pathname":"/docs/zh/ji-chu/unsloth-dynamic-2.0-ggufs","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9a5","description":"我们的 Dynamic Quants 的一次重大新升级！","breadcrumbs":[{"label":"基础"}]},{"id":"9ab45d8a6a66c5a9a735fd4c5d902c7c44b27d87","title":"Aider Polyglot 上的 Unsloth Dynamic GGUF","pathname":"/docs/zh/ji-chu/unsloth-dynamic-2.0-ggufs/unsloth-dynamic-ggufs-on-aider-polyglot","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f9a5","description":"Unsloth Dynamic GGUF 在 Aider Polyglot 基准测试中的表现","breadcrumbs":[{"label":"基础"},{"label":"Unsloth Dynamic 2.0 GGUF","emoji":"1f9a5"}]},{"id":"4fe123c34ab0d523b509efe2b2b56b299498fc5c","title":"本地 LLM 的工具调用指南","pathname":"/docs/zh/ji-chu/tool-calling-guide-for-local-llms","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"screwdriver-wrench","description":"","breadcrumbs":[{"label":"基础"}]},{"id":"5fbd33a15690ccb7e121b10fec182c0112c69066","title":"视觉微调","pathname":"/docs/zh/ji-chu/vision-fine-tuning","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f441","description":"了解如何使用 Unsloth 微调视觉/多模态 LLM","breadcrumbs":[{"label":"基础"}]},{"id":"1e4bf502aabe6ddf3dfea962f3f9d713aaea2190","title":"故障排查与常见问题","pathname":"/docs/zh/ji-chu/troubleshooting-and-faqs","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"26a0","description":"解决问题的技巧，以及常见问题解答。","breadcrumbs":[{"label":"基础"}]},{"id":"d1ca1266e0caf1875e2ec49b943f99476047a4df","title":"Hugging Face Hub，XET 调试","pathname":"/docs/zh/ji-chu/troubleshooting-and-faqs/hugging-face-hub-xet-debugging","siteSpaceId":"sitesp_3cbXc","lang":"zh","description":"调试、排查卡住的下载和缓慢下载","breadcrumbs":[{"label":"基础"},{"label":"故障排查与常见问题","emoji":"26a0"}]},{"id":"5d2a5f8cc953da5c0db5753139d5f3e26a1c4635","title":"聊天模板","pathname":"/docs/zh/ji-chu/chat-templates","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4ac","description":"了解聊天模板的基础知识和自定义选项，包括 Conversational、ChatML、ShareGPT、Alpaca 等格式，以及更多内容！","breadcrumbs":[{"label":"基础"}]},{"id":"5efcf23278a6310dfc435b0e52b441693e4a640d","title":"Unsloth 环境标志","pathname":"/docs/zh/ji-chu/unsloth-environment-flags","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f6e0","description":"如果你看到微调中断，或者想关闭某些功能，这些高级标志可能会很有用。","breadcrumbs":[{"label":"基础"}]},{"id":"eeac0ed07f31cdf24b72252a10d6eceef0bc3355","title":"继续预训练","pathname":"/docs/zh/ji-chu/continued-pretraining","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"267b","description":"也称为持续微调。Unsloth 允许你持续进行预训练，让模型学习新语言。","breadcrumbs":[{"label":"基础"}]},{"id":"bf55f5975c7abaa2eb7d6c35d41d00a5ac7d3046","title":"从最后一个检查点继续微调","pathname":"/docs/zh/ji-chu/finetuning-from-last-checkpoint","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f3c1","description":"检查点保存可让你保存微调进度，以便暂停后继续。","breadcrumbs":[{"label":"基础"}]},{"id":"0faf69cc3c60f48fdf0fcaa450db46623ea2e487","title":"Unsloth 基准测试","pathname":"/docs/zh/ji-chu/unsloth-benchmarks","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"1f4ca","description":"Unsloth 在 NVIDIA GPU 上记录的基准测试。","breadcrumbs":[{"label":"基础"}]},{"id":"124bfded8d8412a9fbc1614fa7467985c0af22da","title":"如何使用 OpenCode 运行本地 AI 模型","pathname":"/docs/zh/ji-cheng/opencode","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"rectangle-vertical","description":"在你的本地设备上将开源 LLM 连接到 OpenCode 的指南。","breadcrumbs":[{"label":"集成"}]},{"id":"f1eb04d9bdae8f6dbb3d9ed5d64e060dac5a68ff","title":"如何使用 OpenClaw 运行本地 AI 模型","pathname":"/docs/zh/ji-cheng/openclaw","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"lobster","description":"使用 OpenClaw 运行本地 LLM 的指南。","breadcrumbs":[{"label":"集成"}]},{"id":"8567c077a06708926cb64d39a91077f5abf44625","title":"如何使用 Hermes Agent 运行本地 AI 模型","pathname":"/docs/zh/ji-cheng/hermes-agent","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"caduceus","description":"关于在本地使用 Hermes Agent 运行开源 LLM 的指南。","breadcrumbs":[{"label":"集成"}]},{"id":"010e01be868ae39c13b48ffdf9774e645c6a347f","title":"将 Python SDK 连接到 Unsloth","pathname":"/docs/zh/ji-cheng/jiang-python-sdk-lian-jie-dao-unsloth","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"python","description":"通过官方 OpenAI 或 Anthropic SDK 从 Python 调用 Unsloth 的本地 API 的指南，包括流式输出、视觉、函数调用，以及 Unsloth 内置的服务器端工具。","breadcrumbs":[{"label":"集成"}]},{"id":"4636d45e7e20328c61211d43c235257fdd7ebc1d","title":"将 Curl 和 HTTP 连接到 Unsloth","pathname":"/docs/zh/ji-cheng/jiang-curl-he-http-lian-jie-dao-unsloth","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"spiral","description":"使用 curl（或任何 HTTP 客户端）调用 Unsloth API 的指南，包含可直接复制粘贴的各端点和功能示例。","breadcrumbs":[{"label":"集成"}]},{"id":"89384e94043ceb248ebc356b1fb9ee69559f1e62","title":"使用 Unsloth Kernels + Packing 实现 3 倍更快的 LLM 训练","pathname":"/docs/zh/bo-ke/3x-faster-training-packing","siteSpaceId":"sitesp_3cbXc","lang":"zh","emoji":"26a1","description":"了解 Unsloth 如何提高训练吞吐量，并消除微调中的 padding 浪费。","breadcrumbs":[{"label":"博客"}]},{"id":"75464ca507ae7bfd1d067969e7f59dd5d631c4e3","title":"50 万上下文长度微调","pathname":"/docs/zh/bo-ke/500k-context-length-fine-tuning","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"ruler-combined","description":"了解如何使用 Unsloth 启用超过 50 万 token 上下文窗口的微调。","breadcrumbs":[{"label":"博客"}]},{"id":"67163c415f870f5b3b00cf5435820df0d5c6e7ce","title":"量化感知训练（QAT）","pathname":"/docs/zh/bo-ke/quantization-aware-training-qat","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"down-left-and-up-right-to-center","description":"使用 Unsloth 和 PyTorch 将模型量化为 4-bit，以恢复准确率。","breadcrumbs":[{"label":"博客"}]},{"id":"702baf22d087bcf50380c3e6f1c28f53999562f2","title":"在 NVIDIA DGX Station 上使用 Unsloth 微调 LLM","pathname":"/docs/zh/bo-ke/dgx-station","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"microchip-ai","description":"关于如何使用 Unsloth 的笔记本在 NVIDIA DGX Station 上进行微调的教程。","breadcrumbs":[{"label":"博客"}]},{"id":"2940860946759c1e409cb13440841f44d539b907","title":"使用 Unsloth 和 Docker 微调 LLM","pathname":"/docs/zh/bo-ke/how-to-fine-tune-llms-with-unsloth-and-docker","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"docker","description":"了解如何使用 Unsloth 的 Docker 镜像微调 LLM 或进行强化学习（RL）。","breadcrumbs":[{"label":"博客"}]},{"id":"abaf2778401f93018feb5d51ff9b83ef161cbc68","title":"使用 NVIDIA DGX Spark 和 Unsloth 微调 LLM","pathname":"/docs/zh/bo-ke/fine-tuning-llms-with-nvidia-dgx-spark-and-unsloth","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"sparkle","description":"关于如何在 NVIDIA DGX Spark 上使用 OpenAI gpt-oss 进行微调和强化学习（RL）的教程。","breadcrumbs":[{"label":"博客"}]},{"id":"402177956bb6f4d27c0bfee66439d5584b91564e","title":"使用 Blackwell、RTX 50 系列与 Unsloth 微调 LLM","pathname":"/docs/zh/bo-ke/fine-tuning-llms-with-blackwell-rtx-50-series-and-unsloth","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"microchip","description":"通过我们的分步指南，学习如何在 NVIDIA 的 Blackwell RTX 50 系列和 B200 GPU 上微调 LLM。","breadcrumbs":[{"label":"博客"}]},{"id":"af798f5c75ff3a274d4f221dc715d88c6daf3870","title":"释放 AMD 的力量：Unsloth 官方支持现已上线！","pathname":"/docs/zh/bo-ke/shi-fang-amd-de-li-liang-unsloth-guan-fang-zhi-chi-xian-yi-shang-xian","siteSpaceId":"sitesp_3cbXc","lang":"zh","icon":"square-up-right","description":"Unsloth 对 AMD GPU 的支持现已正式上线。无需 NVIDIA 硬件，即可将 LLM 微调速度提升至 2 倍，内存占用约减少 70%。","breadcrumbs":[{"label":"博客"}]},{"id":"6e2677498ce15bfc11f46d16a61c61c1a7ff3edd","title":"Unslothドキュメント","pathname":"/docs/jp","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9a5","description":"Unslothは、モデルの実行と学習のためのオープンソースフレームワークです。","breadcrumbs":[{"label":"始める"}]},{"id":"f10fbc893d08889a92e28bfa2fc8fe09c696363d","title":"初心者向けファインチューニング","pathname":"/docs/jp/meru/fine-tuning-for-beginners","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"2b50","description":"","breadcrumbs":[{"label":"始める"}]},{"id":"6d173d71b33fc6bb9bbdbeb7a340a80132236305","title":"Unslothの要件","pathname":"/docs/jp/meru/fine-tuning-for-beginners/unsloth-requirements","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f6e0","description":"こちらは、システム要件とGPU VRAM要件を含むUnslothの要件です。","breadcrumbs":[{"label":"始める"},{"label":"初心者向けファインチューニング","emoji":"2b50"}]},{"id":"fffb67e71b118ce93cf382a53722ac7a1a41dd7c","title":"FAQ + ファインチューニングは私に向いていますか？","pathname":"/docs/jp/meru/fine-tuning-for-beginners/faq-+-is-fine-tuning-right-for-me","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f914","description":"ファインチューニングが自分に合っているか迷っているなら、こちらをご覧ください！ファインチューニングの誤解、RAGとの比較などを学べます:","breadcrumbs":[{"label":"始める"},{"label":"初心者向けファインチューニング","emoji":"2b50"}]},{"id":"a3325bc180768721f1119499c1b82e6d84e53794","title":"Unslothノートブック","pathname":"/docs/jp/meru/unsloth-notebooks","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4d2","description":"ファインチューニング用ノートブック: Unslothカタログを見てみましょう。","breadcrumbs":[{"label":"始める"}]},{"id":"189a8deb3aa7eabc5be5a7563f26d4b08f252e28","title":"Unslothモデルカタログ","pathname":"/docs/jp/meru/unsloth-model-catalog","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f52e","description":"","breadcrumbs":[{"label":"始める"}]},{"id":"78ed2140c0094b3caa1931c2127e336118049b42","title":"Unslothのインストール","pathname":"/docs/jp/meru/install","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4e5","description":"Unslothをローカルまたはオンラインでインストールする方法を学びましょう。","breadcrumbs":[{"label":"始める"}]},{"id":"2668025f87c0405c2cc5a80af272adcb7670b1c1","title":"pipとuvでUnslothをインストール","pathname":"/docs/jp/meru/install/pip-install","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"desktop-arrow-down","description":"Pipを使ってローカルにUnslothをインストールするには、以下の手順に従ってください:","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"f0aec71b27dd3fb7098c85a48ee39b11e05798fd","title":"MacOSにUnslothをインストール","pathname":"/docs/jp/meru/install/mac","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"apple","description":"","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"3f25ebdeb2b687d90421588e1b9a97b6f91995fe","title":"Unslothを使ってWindowsでLLMをファインチューニングする方法（ステップバイステップガイド）","pathname":"/docs/jp/meru/install/windows-installation","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"windows","description":"ローカルでLLMのファインチューニングを始めるために、WindowsへのUnslothのインストール方法を確認しましょう。","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"c188f95d26bbe239b9750e6595c7164791be0287","title":"DockerでUnslothをインストール","pathname":"/docs/jp/meru/install/docker","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"docker","description":"公式Dockerコンテナを使ってUnslothをインストールします","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"a7e24b56702a0f7fb45f9f2ce0e6bd34dd1d2f50","title":"Unslothの更新","pathname":"/docs/jp/meru/install/updating","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"arrow-rotate-right","description":"Unslothを更新する、または古いバージョンを使うには、以下の手順に従ってください:","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"cc479234476bfcb2928aaa70e0b137d2f1b29981","title":"UnslothガイドでAMD GPU上のLLMをファインチューニング","pathname":"/docs/jp/meru/install/amd","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"square-up-right","description":"Unslothを使ってAMD GPU上で大規模言語モデル（LLM）をファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"69a2c17173ec9337f4907132dee8c77a4517814a","title":"Unslothを使ってIntel GPU上のLLMをファインチューニング","pathname":"/docs/jp/meru/install/intel","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"info","description":"Intel GPU上で大規模言語モデルを学習・ファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"始める"},{"label":"Unslothのインストール","emoji":"1f4e5"}]},{"id":"ae7926a0f371a1f594e3344d85571edf864a5259","title":"LLMファインチューニングガイド","pathname":"/docs/jp/meru/fine-tuning-llms-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9ec","description":"ファインチューニングの基本とベストプラクティスをすべて学びましょう。初心者向けです。","breadcrumbs":[{"label":"始める"}]},{"id":"0169f16f28bd29579ad4805156b35641f23090aa","title":"データセットガイド","pathname":"/docs/jp/meru/fine-tuning-llms-guide/datasets-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4c8","description":"ファインチューニング用データセットの作成と準備方法を学びましょう。","breadcrumbs":[{"label":"始める"},{"label":"LLMファインチューニングガイド","emoji":"1f9ec"}]},{"id":"dac5c97d7a208c227b8115ec72133012ad2d0113","title":"LoRAファインチューニングのハイパーパラメータガイド","pathname":"/docs/jp/meru/fine-tuning-llms-guide/lora-hyperparameters-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9e0","description":"最適なLLMファインチューニング設定をステップバイステップで学びましょう。LoRAのランクとalpha、エポック、バッチサイズ＋勾配蓄積、QLoRA対LoRA、対象モジュールなどを扱います。","breadcrumbs":[{"label":"始める"},{"label":"LLMファインチューニングガイド","emoji":"1f9ec"}]},{"id":"b0eaf2528a8f433d814b471834923a39c77f801f","title":"どのモデルをファインチューニングに使うべきですか？","pathname":"/docs/jp/meru/fine-tuning-llms-guide/what-model-should-i-use","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"2753","description":"","breadcrumbs":[{"label":"始める"},{"label":"LLMファインチューニングガイド","emoji":"1f9ec"}]},{"id":"f46773a93603b1da44e6319f55fd3b900d7700fc","title":"チュートリアル: Llama-3をファインチューニングしてOllamaで使う方法","pathname":"/docs/jp/meru/fine-tuning-llms-guide/tutorial-how-to-finetune-llama-3-and-use-in-ollama","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f999","description":"Ollama上でローカル実行するカスタマイズされた個人アシスタント（ChatGPTのようなもの）を作成する初心者向けガイド","breadcrumbs":[{"label":"始める"},{"label":"LLMファインチューニングガイド","emoji":"1f9ec"}]},{"id":"2e1ff161b8dd839d98df40642b2647753c4a80e4","title":"強化学習（RL）ガイド","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4a1","description":"強化学習（RL）のすべてと、GRPOを使ってUnslothで自分のDeepSeek-R1推論モデルを学習する方法を学びましょう。初心者から上級者まで対応した完全ガイドです。","breadcrumbs":[{"label":"始める"}]},{"id":"7e5df0bd00dfa415f0af9c38f10687153fb5f588","title":"7倍長いコンテキストでの強化学習GRPO","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/grpo-long-context","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f300","description":"Unslothがどのように超長文コンテキストのRLファインチューニングを可能にするかを学びましょう。","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"01cdfed2f29f2821ff712fcb0ccb837f8ba4baf3","title":"ビジョン強化学習（VLM RL）","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/vision-reinforcement-learning-vlm-rl","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f441-1f5e8","description":"Unslothを使ってGRPOとRLでビジョン/マルチモーダルモデルを学習しましょう！","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"fb9d3ff64a5035fc434979bb3b39a92149730468","title":"FP8強化学習","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/fp8-reinforcement-learning","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f3b1","description":"UnslothでFP8精度による強化学習（RL）とGRPOを学習しましょう。","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"db5a7a83b913f48723eb2d7a67636b192c105cd3","title":"チュートリアル: GRPOで自分の推論モデルを学習する","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/tutorial-train-your-own-reasoning-model-with-grpo","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"26a1","description":"UnslothとGRPOを使って、Llama 3.1（8B）のようなモデルを推論モデルへ変換する初心者向けガイド。","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"d162fad71aa9478e713b7a7dfa3d60267dfca53b","title":"高度な強化学習ドキュメント","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/advanced-rl-documentation","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9e9","description":"UnslothをGRPOと併用する際の高度なドキュメント設定。","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"a803c7bb314b96d0901b440bc199389cb1bd3663","title":"GSPO強化学習","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/advanced-rl-documentation/gspo-reinforcement-learning","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"lightbulb-on","description":"UnslothでGSPO（Group Sequence Policy Optimization）RLを使って学習します。","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"},{"label":"高度な強化学習ドキュメント","emoji":"1f9e9"}]},{"id":"a653c474327e232378ecbe04adeeaf5e3b6abaaa","title":"RL報酬ハッキング","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/advanced-rl-documentation/rl-reward-hacking","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"treasure-chest","description":"強化学習における報酬ハッキングとは何か、そしてそれを防ぐ方法を学びましょう。","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"},{"label":"高度な強化学習ドキュメント","emoji":"1f9e9"}]},{"id":"279889a232e1a4ec554827f0eb7c954d2871a4a3","title":"RLにおけるFP16とBF16","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/advanced-rl-documentation/fp16-vs-bf16-for-rl","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"2049","description":"「Defeating the Training-Inference Mismatch via FP16 https://arxiv.org/pdf/2510.26788」では、float16の使用がbfloat16より優れていることが示されています","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"},{"label":"高度な強化学習ドキュメント","emoji":"1f9e9"}]},{"id":"db44fb02e3caa0e241e41eb5c751f1554262cfb3","title":"メモリ効率の高いRL","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/memory-efficient-rl","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"memory","description":"","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"7f735ce67d58d1d8b36af1a654b51e99d8461a8f","title":"選好最適化学習 - DPO、ORPO、KTO","pathname":"/docs/jp/meru/reinforcement-learning-rl-guide/preference-dpo-orpo-and-kto","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f3c6","description":"Unslothを使ってDPO、GRPO、ORPO、KTOによる選好アライメントのファインチューニングを学びましょう。以下の手順に従ってください:","breadcrumbs":[{"label":"始める"},{"label":"強化学習（RL）ガイド","emoji":"1f4a1"}]},{"id":"a01cba7455dcbbf4a34e17600cd0b0101df41d37","title":"Unsloth Studioの紹介","pathname":"/docs/jp/xin-zhe/studio","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9a5","description":"Unsloth Studioを使ってAIモデルをローカルで実行・学習しましょう。","breadcrumbs":[{"label":"新着"}]},{"id":"511d3548745f95b53d7c4df809a4aeb55f219840","title":"Unsloth Studioを始める","pathname":"/docs/jp/xin-zhe/studio/start","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"bolt","description":"ファインチューニングスタジオ、データレシピ、モデルのエクスポート、チャットを始めるためのガイド。","breadcrumbs":[{"label":"新着"},{"label":"Unsloth Studioの紹介","emoji":"1f9a5"}]},{"id":"ca5ab7d9f589ffe0771972c3c3b4665342eb3baf","title":"Unsloth Studioでモデルを実行する方法","pathname":"/docs/jp/xin-zhe/studio/chat","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"comment-dots","description":"Unsloth Studioを使ってAIモデル、LLM、GGUFをローカルで実行します。","breadcrumbs":[{"label":"新着"},{"label":"Unsloth Studioの紹介","emoji":"1f9a5"}]},{"id":"8454803abb981b29f6c46027668d1416d8da0199","title":"Unsloth Studioのインストール","pathname":"/docs/jp/xin-zhe/studio/install","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"arrow-down-to-square","description":"ローカルデバイスにUnsloth Studioをインストールする方法を学びましょう。","breadcrumbs":[{"label":"新着"},{"label":"Unsloth Studioの紹介","emoji":"1f9a5"}]},{"id":"1efc27ff794b527bf40ca1399ab10a6c28cc4b61","title":"Unslothデータレシピ","pathname":"/docs/jp/xin-zhe/studio/data-recipe","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"hat-chef","description":"Unsloth StudioのData Recipesを使ってデータセットを作成、構築、編集する方法を学びましょう。","breadcrumbs":[{"label":"新着"},{"label":"Unsloth Studioの紹介","emoji":"1f9a5"}]},{"id":"a58a8ab897451539e1493312c6a640b4d5ee40b7","title":"Unsloth Studioでモデルをエクスポートする","pathname":"/docs/jp/xin-zhe/studio/export","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"box-isometric","description":"safetensorやLoRAのモデルファイルをGGUFや他の形式にエクスポートする方法を学びましょう。","breadcrumbs":[{"label":"新着"},{"label":"Unsloth Studioの紹介","emoji":"1f9a5"}]},{"id":"8a5c4db36cff597fc13ab7546333ece742a1f076","title":"Unslothの更新情報","pathname":"/docs/jp/xin-zhe/changelog","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"sparkles","description":"最新リリース、改善、修正に関するUnslothの変更履歴。","breadcrumbs":[{"label":"新着"}]},{"id":"b1345d93eb2f70a681e5a0bf731e28a576f0f428","title":"Qwen3.6 - ローカル実行方法","pathname":"/docs/jp/moderu/qwen3.6","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f49c","description":"新しいQwen3.6-27Bと35B-A3Bモデルをローカルで実行しましょう！","breadcrumbs":[{"label":"モデル"}]},{"id":"693bc7a2f22dcaf0c6bc0818f2076196fe331fa7","title":"Gemma 4 - ローカル実行方法","pathname":"/docs/jp/moderu/gemma-4","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"2728","description":"Googleの新しいGemma 4モデルを、E2B、E4B、26B A4B、31Bを含めてローカルで実行しましょう。","breadcrumbs":[{"label":"モデル"}]},{"id":"4a6e7bbec569d341f876db55593564610de4d0a8","title":"Gemma 4ファインチューニングガイド","pathname":"/docs/jp/moderu/gemma-4/train","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"flask-gear","description":"GoogleのGemma 4をUnslothで学習しましょう。","breadcrumbs":[{"label":"モデル"},{"label":"Gemma 4 - ローカル実行方法","emoji":"2728"}]},{"id":"3b841dc831ba1d1cd9a579d65ec4951bca9c1e85","title":"NVIDIA Nemotron 3 Nano Omni - ローカル実行方法","pathname":"/docs/jp/moderu/nemotron-3-nano-omni","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9e9","description":"Nemotron-3-Nano-Omni-30B-A3Bをローカルデバイスで実行・ファインチューニングしましょう！","breadcrumbs":[{"label":"モデル"}]},{"id":"ba28d5a94414d9422f453043c0cb720721211a60","title":"Kimi K2.6 - ローカル実行方法","pathname":"/docs/jp/moderu/kimi-k2.6","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f95d","description":"Kimi-K2.6を自分のローカルデバイスで実行するステップバイステップガイド。","breadcrumbs":[{"label":"モデル"}]},{"id":"0af04e20683a2825742edd360e0a15913f42c5a8","title":"Qwen3.5 - ローカル実行方法","pathname":"/docs/jp/moderu/qwen3.5","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f49c","description":"新しいQwen3.5 LLMを、Medium: Qwen3.5-35B-A3B、27B、122B-A10B、Small: Qwen3.5-0.8B、2B、4B、9B、397B-A17B を含めてローカルデバイスで実行しましょう！","breadcrumbs":[{"label":"モデル"}]},{"id":"a38c8e31b4301f0ec111df120e156fefb092bbd8","title":"Qwen3.5ファインチューニングガイド","pathname":"/docs/jp/moderu/qwen3.5/fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"flask-gear","description":"Unslothを使ってQwen3.5 LLMをファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"モデル"},{"label":"Qwen3.5 - ローカル実行方法","emoji":"1f49c"}]},{"id":"bb7af419ae482d9b8efadbd19e33855d14ce5ee9","title":"Qwen3.5 GGUFベンチマーク","pathname":"/docs/jp/moderu/qwen3.5/gguf-benchmarks","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"chart-fft","description":"Unsloth Dynamic GGUFの性能と、perplexity、KL divergence、MXFP4の分析をご覧ください。","breadcrumbs":[{"label":"モデル"},{"label":"Qwen3.5 - ローカル実行方法","emoji":"1f49c"}]},{"id":"bbfe7c6edcb58685ec7cf51092f2663c6c435010","title":"GLM-5.1 - ローカル実行方法","pathname":"/docs/jp/moderu/glm-5.1","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"z","description":"Z.aiの新しいGLM-5.1モデルを自分のローカルデバイスで実行しましょう！","breadcrumbs":[{"label":"モデル"}]},{"id":"ea13ebdc099377f5b37a2e0895aeb9ede74953ed","title":"大規模言語モデル（LLM）チュートリアル","pathname":"/docs/jp/moderu/tutorials","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f680","description":"","breadcrumbs":[{"label":"モデル"}]},{"id":"def6c4e623d8ca4355a9f45aa124675d9b1bc37e","title":"Qwen3 - 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/qwen3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f320","description":"UnslothとDynamic 2.0量子化を使ってQwen3をローカルで実行・ファインチューニングする方法を学びましょう","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"41930a9c7687512da264c245376cdad995638965","title":"Qwen3-VL: 実行ガイド","pathname":"/docs/jp/moderu/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-vl-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f320","description":"Unslothを使ってQwen3-VLをローカルでファインチューニング・実行する方法を学びましょう。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"},{"label":"Qwen3 - 実行とファインチューニング方法","emoji":"1f320"}]},{"id":"fa5bc8c4a776b4aa07f35a8be3714e534db69c0d","title":"Qwen3-2507: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-2507","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f320","description":"Qwen3-30B-A3B-2507 と 235B-A22B の Thinking版と Instruct版をデバイス上でローカル実行しましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"},{"label":"Qwen3 - 実行とファインチューニング方法","emoji":"1f320"}]},{"id":"30aa264fb1374e73acfe9203ecfdbd4556a9335b","title":"MiniMax-M2.7 - ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/minimax-m27","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"waveform","description":"MiniMax-M2.7 LLMを自分のデバイスでローカル実行しましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"5d12825037ce6a6f7dbf0360c12a373f136dfaf1","title":"GLM-5: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/glm-5","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"z","description":"Z.aiの新しいGLM-5モデルを自分のローカルデバイスで実行しましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"17c4f4f0107699d7b64bfeb4b34d427df32a68f9","title":"Kimi K2.5: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/kimi-k2.5","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f95d","description":"Kimi-K2.5を自分のローカルデバイスで実行するガイド！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"68cc5d489fe86eefb1bb9cd2e351bcdb90866b49","title":"GLM-4.7-Flash: ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/glm-4.7-flash","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"z","description":"GLM-4.7-Flashをデバイス上でローカル実行・ファインチューニングしましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"051e934a3c33af7cc4b8bd92b687148f8f65e39e","title":"MiniMax-M2.5: 実行ガイド","pathname":"/docs/jp/moderu/tutorials/minimax-m25","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"waveform","description":"MiniMax-M2.5を自分のデバイスでローカル実行しましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"14947b7a6d6c50557f1e971fcfb24aa4e19eeb01","title":"Qwen3-Coder: ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/qwen3-coder-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f320","description":"Unsloth Dynamic量子化を使ってQwen3-Coder-30B-A3B-Instruct と 480B-A35B をローカルで実行しましょう。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"6248d03dd6d5a8cecb057bb4c904aa6596097dcb","title":"Gemma 3 - 実行ガイド","pathname":"/docs/jp/moderu/tutorials/gemma-3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"google","description":"llama.cpp、Ollama、Open WebUI上で当社のGGUFを使ってGemma 3を効果的に実行する方法と、Unslothでファインチューニングする方法を紹介します！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"51b993bab9cf405fd074014ca3dba97ff1a1ac56","title":"Gemma 3n: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/gemma-3-how-to-run-and-fine-tune/gemma-3n-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"google","description":"Googleの新しいGemma 3nを、llama.cpp、Ollama、Open WebUI上でDynamic GGUFを使ってローカル実行し、Unslothでファインチューニングしましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"},{"label":"Gemma 3 - 実行ガイド","icon":"google"}]},{"id":"f7d04e17a76834c8a0c837ec4aee9d3d64c1c9db","title":"DeepSeek-OCR 2: 実行とファインチューニングガイド","pathname":"/docs/jp/moderu/tutorials/deepseek-ocr-2","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f433","description":"DeepSeek-OCR-2をローカルで実行・ファインチューニングする方法のガイド。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"d120eefd9dfbf755a96bf2cab5ea0bef3bb5e508","title":"GLM-4.7: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/glm-4.7","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"z","description":"Z.aiのGLM-4.7モデルを自分のローカルデバイスで実行する方法のガイド！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"a3a6dc010c538ced0d9db88c169b13882b595c54","title":"ComfyUIでQwen-Image-2512をローカル実行する方法","pathname":"/docs/jp/moderu/tutorials/qwen-image-2512","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f49f","description":"ComfyUIを使ってQwen-Image-2512をローカルデバイスで実行するためのステップバイステップチュートリアル。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"1b1608b9402b46fbfed79b5defc616e893115b63","title":"stable-diffusion.cppでQwen-Image-2512を実行するチュートリアル","pathname":"/docs/jp/moderu/tutorials/qwen-image-2512/stable-diffusion.cpp","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f3a8","description":"stable-diffusion.cppでQwen-Image-2512を使うためのチュートリアル。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"},{"label":"ComfyUIでQwen-Image-2512をローカル実行する方法","emoji":"1f49f"}]},{"id":"6e250b10672043df657699c97a0ef9185dd3fcf8","title":"Devstral 2 - 実行ガイド","pathname":"/docs/jp/moderu/tutorials/devstral-2","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4d9","description":"Mistral Devstral 2モデル（123B-Instruct-2512とSmall-2-24B-Instruct-2512）をローカルで実行するためのガイド。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"d09407ea18ce3f2ae1895fad1fea42fea44a7084","title":"Ministral 3 - 実行ガイド","pathname":"/docs/jp/moderu/tutorials/ministral-3","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f431","description":"Mistral Ministral 3モデルを、ローカルデバイスで実行またはファインチューニングするためのガイド","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"f6ca689c1f955e7e2dc3b6b42bebc940f63a0c1a","title":"DeepSeek-OCR: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/deepseek-ocr-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f433","description":"DeepSeek-OCRをローカルで実行・ファインチューニングする方法のガイド。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"d33b81014bee91e7c28cf5f645d19f717568e403","title":"Kimi K2 Thinking: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/kimi-k2-thinking-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f319","description":"Kimi-K2-ThinkingとKimi-K2を自分のローカルデバイスで実行するためのガイド！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"15da73ffae68627da854f843c0328406ea1ae99b","title":"GLM-4.6: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/glm-4.6-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"z","description":"Z.aiのGLM-4.6およびGLM-4.6V-Flashモデルを自分のローカルデバイスで実行する方法のガイド！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"b2a04bf177dc4438c58a13a8bff973ff4b9d1cf8","title":"Qwen3-Next: ローカル実行ガイド","pathname":"/docs/jp/moderu/tutorials/qwen3-next","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f320","description":"Qwen3-Next-80B-A3B-Instruct と Thinking版をデバイス上でローカル実行しましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"c590d523e85962de32ad10f332254fb7ec5aa904","title":"FunctionGemma: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/functiongemma","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"google","description":"FunctionGemmaを自分のデバイスやスマホでローカル実行・ファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"1076a52a5c3ebf51b7a9fff0968fcea31aee4df1","title":"DeepSeek-V3.1: ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/deepseek-v3.1-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f40b","description":"DeepSeek-V3.1とTerminusを自分のローカルデバイスで実行する方法のガイド！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"25a2638b85cdb3aec85d94e5705fb8ee19a1daab","title":"DeepSeek-R1-0528: ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/deepseek-r1-0528-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f40b","description":"Qwen3を含むDeepSeek-R1-0528を自分のローカルデバイスで実行する方法のガイド！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"56706e6412f8aee419d6c7850e9d0a63951b6c8f","title":"Liquid LFM2.5: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/lfm2.5","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4a7","description":"LFM2.5 InstructとVisionをデバイス上でローカル実行・ファインチューニングしましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"8df1f0c90d5752fe77400c0eb7e1397df03d993c","title":"Magistral: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/magistral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4a5","description":"Magistralへようこそ - Mistralの新しい推論モデルです。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"ea8d2ac599fb9a6e666fafcc4696434ad323e8ef","title":"IBM Granite 4.0","pathname":"/docs/jp/moderu/tutorials/ibm-granite-4.0","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"cube","description":"llama.cpp、Ollama上でUnsloth GGUFを使ってIBM Granite-4.0を実行する方法と、ファインチューニングの方法！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"49c5cefd40453c244bfe13e6bac549a8d476c3c8","title":"Llama 4: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/llama-4-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f999","description":"標準的な量子化よりも精度を回復できる当社のDynamic GGUFを使って、Llama 4をローカルで実行する方法。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"4e9f2261069b3d110ae7851ab15866b1a06c1a2a","title":"Grok 2","pathname":"/docs/jp/moderu/tutorials/grok-2","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"square-x-twitter","description":"xAIのGrok 2モデルをローカルで実行しましょう！","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"31e0b99ff41cc424049d2d4d44d650fb271d0392","title":"Devstral: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/devstral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4d9","description":"Small-2507と2505を含むMistral Devstral 1.1を実行・ファインチューニングしましょう。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"b00c9977bf9de6379716fbcd59dde91df9f3e609","title":"DockerでローカルLLMを実行する方法: ステップバイステップガイド","pathname":"/docs/jp/moderu/tutorials/how-to-run-llms-with-docker","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"docker","description":"DockerとUnslothを使ってローカルデバイスで大規模言語モデル（LLM）を実行する方法を学びましょう。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"762541362c9823bb63ea9e0ca527832550a99694","title":"DeepSeek-V3-0324: ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/deepseek-v3-0324-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f433","description":"精度を回復する当社のDynamic量子化を使ってDeepSeek-V3-0324をローカルで実行する方法","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"551e9305c991b8df9f92f34988ae3d56b2fd8357","title":"DeepSeek-R1: ローカル実行方法","pathname":"/docs/jp/moderu/tutorials/deepseek-r1-how-to-run-locally","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f40b","description":"llama.cppを使ってDeepSeek-R1向けの1.58-bit Dynamic Quantsを実行する方法のガイド。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"178cfb29f5efd456d045bfa2bb771ca13cb6b8ed","title":"DeepSeek-R1 Dynamic 1.58-bit","pathname":"/docs/jp/moderu/tutorials/deepseek-r1-how-to-run-locally/deepseek-r1-dynamic-1.58-bit","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f433","description":"UnslothのDynamic GGUF Quantsと標準IMatrix Quantsの性能比較表をご覧ください。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"},{"label":"DeepSeek-R1: ローカル実行方法","emoji":"1f40b"}]},{"id":"9fceb033f7114c1fe5f72941ba83dc6ec92b8a1d","title":"Phi-4 Reasoning: 実行とファインチューニング方法","pathname":"/docs/jp/moderu/tutorials/phi-4-reasoning-how-to-run-and-fine-tune","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"windows","description":"UnslothとDynamic 2.0量子化を使ってPhi-4 reasoningモデルをローカルで実行・ファインチューニングする方法を学びましょう","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"d22cc824eafc19900f391f15f9dc0e258f4479a3","title":"QwQ-32B: 効果的な実行方法","pathname":"/docs/jp/moderu/tutorials/qwq-32b-how-to-run-effectively","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f320","description":"当社の修正版と、無限生成を避ける設定、GGUFを使ってQwQ-32Bを効果的に実行する方法。","breadcrumbs":[{"label":"モデル"},{"label":"大規模言語モデル（LLM）チュートリアル","emoji":"1f680"}]},{"id":"664bab64c0c5cdbe1a11fa3ba21ffe4c16cf0f52","title":"UnslothをAPIエンドポイントとして使う方法","pathname":"/docs/jp/ji-ben/api","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"globe","description":"","breadcrumbs":[{"label":"基本"}]},{"id":"d5ae43f1915ceda3d304ad7c413cb4efbe3d1a3f","title":"推論とデプロイ","pathname":"/docs/jp/ji-ben/inference-and-deployment","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f5a5","description":"ファインチューニング済みモデルを保存して、お気に入りの推論エンジンで実行できるようにする方法を学びましょう。","breadcrumbs":[{"label":"基本"}]},{"id":"9bfa988baa17c249340a58c332b8584f20d2537c","title":"GGUFへの保存","pathname":"/docs/jp/ji-ben/inference-and-deployment/saving-to-gguf","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"16a93a7cb5d8a8ae32bc84526191a967ae25818a","title":"Speculative Decoding","pathname":"/docs/jp/ji-ben/inference-and-deployment/saving-to-gguf/speculative-decoding","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"llama-server、llama.cpp、vLLMなどを使ったSpeculative Decodingで、推論を2倍高速化","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"},{"label":"GGUFへの保存"}]},{"id":"0fde417d83989a8108b1d466ec2b53c46e9f4279","title":"vLLMデプロイメントと推論ガイド","pathname":"/docs/jp/ji-ben/inference-and-deployment/vllm-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"本番環境でLLMを提供するために、LLMをvLLMへ保存・デプロイするためのガイド","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"c81cf073f3b0e1fc8edcdbc81171084f13fcd40a","title":"vLLMエンジン引数","pathname":"/docs/jp/ji-ben/inference-and-deployment/vllm-guide/vllm-engine-arguments","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"},{"label":"vLLMデプロイメントと推論ガイド"}]},{"id":"087d5c3a4a847726fc6a5174ab1fd71832977239","title":"LoRAホットスワッピングガイド","pathname":"/docs/jp/ji-ben/inference-and-deployment/vllm-guide/lora-hot-swapping-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"},{"label":"vLLMデプロイメントと推論ガイド"}]},{"id":"52955f68c3f0eb3d6d351166737e23bf6d88b360","title":"Ollamaへの保存","pathname":"/docs/jp/ji-ben/inference-and-deployment/saving-to-ollama","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"6fb5e7690fd265f7cbc75a7c89a22f8e97491f6e","title":"LM Studioへのモデルデプロイ","pathname":"/docs/jp/ji-ben/inference-and-deployment/lm-studio","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"LM Studioで実行・デプロイできるように、モデルをGGUFへ保存する方法","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"b192549682c1e2b38258e6cf4ecfac5d454f5919","title":"LinuxターミナルにLM Studio CLIをインストールする方法","pathname":"/docs/jp/ji-ben/inference-and-deployment/lm-studio/how-to-install-lm-studio-cli-in-linux-terminal","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f47e","description":"ターミナル上でUIなしでLM Studio CLIをインストールするガイド。","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"},{"label":"LM Studioへのモデルデプロイ"}]},{"id":"1f909a31e96a93e6296ea02f561cefd3bfad5cfb","title":"SGLangデプロイメントと推論ガイド","pathname":"/docs/jp/ji-ben/inference-and-deployment/sglang-guide","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"本番環境でLLMを提供するために、LLMをSGLangへ保存・デプロイするためのガイド","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"bf15adf206552e2848b9122be6f17e403b1fc9c0","title":"llama-serverとOpenAIエンドポイントのデプロイガイド","pathname":"/docs/jp/ji-ben/inference-and-deployment/llama-server-and-openai-endpoint","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"OpenAI互換エンドポイントを備えたllama-server経由でのデプロイ","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"6b439ad8eedcbf85fcbca5c3aa910e0d71794507","title":"iOSまたはAndroidスマホでLLMを実行・デプロイする方法","pathname":"/docs/jp/ji-ben/inference-and-deployment/deploy-llms-phone","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4f1","description":"自分のLLMをファインチューニングし、ExecuTorchを使ってAndroidまたはiPhoneにデプロイするチュートリアル。","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"d7914a52b10ea48e9158d90873a1008f60c5f306","title":"推論のトラブルシューティング","pathname":"/docs/jp/ji-ben/inference-and-deployment/troubleshooting-inference","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"モデルの実行や保存で問題が発生している場合。","breadcrumbs":[{"label":"基本"},{"label":"推論とデプロイ","emoji":"1f5a5"}]},{"id":"ee610b22aa43d29d8415fd27eb7de15ba88f7385","title":"Claude CodeでローカルLLMを実行する方法","pathname":"/docs/jp/ji-ben/claude-code","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"claude","description":"ローカルデバイスでClaude Codeとオープンモデルを使うためのガイド。","breadcrumbs":[{"label":"基本"}]},{"id":"c87896ff7159620f4c01bb39fe9df1fd1a55274e","title":"OpenAI CodexでローカルLLMを実行する方法","pathname":"/docs/jp/ji-ben/codex","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"openai","description":"デバイス上でOpenAI Codexとオープンモデルをローカルで使います。","breadcrumbs":[{"label":"基本"}]},{"id":"2cb214340788afbb22dd3dbc72f5295f646274e5","title":"UnslothによるマルチGPUファインチューニング","pathname":"/docs/jp/ji-ben/multi-gpu-training-with-unsloth","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"rectangle-history","description":"Unslothを使って複数GPUと並列処理でLLMをファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"基本"}]},{"id":"76cc4270e39e2817549ae7d884b3b38c0e064b48","title":"Distributed Data Parallel（DDP）によるマルチGPUファインチューニング","pathname":"/docs/jp/ji-ben/multi-gpu-training-with-unsloth/ddp","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"Unsloth CLIを使ってDistributed Data Parallel（DDP）で複数GPU学習を行う方法を学びましょう！","breadcrumbs":[{"label":"基本"},{"label":"UnslothによるマルチGPUファインチューニング","icon":"rectangle-history"}]},{"id":"a637d2257857c4677688add9f50f851ec564d0b8","title":"UnslothガイドによるEmbeddingモデルのファインチューニング","pathname":"/docs/jp/ji-ben/embedding-finetuning","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f50e","description":"Unslothを使ってEmbeddingモデルを簡単にファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"基本"}]},{"id":"084af8821a599e81fc9cc18b65615343a785fc5d","title":"UnslothでMoEモデルを12倍高速にファインチューニング","pathname":"/docs/jp/ji-ben/faster-moe","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f48e","description":"Unslothガイドを使ってMoE LLMをローカルで学習しましょう。","breadcrumbs":[{"label":"基本"}]},{"id":"abcb971ab51a2b38849d11fcae86c3a06fbca382","title":"Text-to-Speech（TTS）ファインチューニングガイド","pathname":"/docs/jp/ji-ben/text-to-speech-tts-fine-tuning","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f50a","description":"Unslothを使ってTTSとSTTの音声モデルをファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"基本"}]},{"id":"021c19d75a86d940ef9b83b7b41bee381657345b","title":"Unsloth Dynamic 2.0 GGUF","pathname":"/docs/jp/ji-ben/unsloth-dynamic-2.0-ggufs","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9a5","description":"Dynamic Quantsの大きな新アップグレードです！","breadcrumbs":[{"label":"基本"}]},{"id":"bef806b79aad8a62f0bba5a87c182261ad394be4","title":"Aider Polyglot上のUnsloth Dynamic GGUF","pathname":"/docs/jp/ji-ben/unsloth-dynamic-2.0-ggufs/unsloth-dynamic-ggufs-on-aider-polyglot","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f9a5","description":"Aider PolyglotベンチマークにおけるUnsloth Dynamic GGUFの性能","breadcrumbs":[{"label":"基本"},{"label":"Unsloth Dynamic 2.0 GGUF","emoji":"1f9a5"}]},{"id":"34b68a962f8c73059943abdee91772400b1d1ecb","title":"ローカルLLMのツール呼び出しガイド","pathname":"/docs/jp/ji-ben/tool-calling-guide-for-local-llms","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"screwdriver-wrench","description":"","breadcrumbs":[{"label":"基本"}]},{"id":"4c4214eb859e3837d4cc099505565e44dba70ac2","title":"ビジョンファインチューニング","pathname":"/docs/jp/ji-ben/vision-fine-tuning","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f441","description":"Unslothを使ってビジョン/マルチモーダルLLMをファインチューニングする方法を学びましょう","breadcrumbs":[{"label":"基本"}]},{"id":"f2407144d3c7f3017e630580e779721eb687c843","title":"トラブルシューティングとFAQ","pathname":"/docs/jp/ji-ben/troubleshooting-and-faqs","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"26a0","description":"問題を解決するためのヒントと、よくある質問。","breadcrumbs":[{"label":"基本"}]},{"id":"87c1a7628195f3ad3d84a0f7d73d266d9af883f1","title":"Hugging Face Hub、XETデバッグ","pathname":"/docs/jp/ji-ben/troubleshooting-and-faqs/hugging-face-hub-xet-debugging","siteSpaceId":"sitesp_8AL84","lang":"ja","description":"デバッグ、停止したダウンロードのトラブルシューティング、遅いダウンロード","breadcrumbs":[{"label":"基本"},{"label":"トラブルシューティングとFAQ","emoji":"26a0"}]},{"id":"775aacfac2bd0502888931ca80934e75342ef4f6","title":"チャットテンプレート","pathname":"/docs/jp/ji-ben/chat-templates","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4ac","description":"会話形式、ChatML、ShareGPT、Alpaca形式などを含む、チャットテンプレートの基本とカスタマイズオプションを学びましょう！","breadcrumbs":[{"label":"基本"}]},{"id":"05baecaeb43111cac111ccadefa0a1e6b0c7cc6e","title":"Unsloth環境フラグ","pathname":"/docs/jp/ji-ben/unsloth-environment-flags","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f6e0","description":"ファインチューニングが壊れるのを見かけたときや、何かを無効化したいときに役立つ高度なフラグ。","breadcrumbs":[{"label":"基本"}]},{"id":"ad4913741729d4d5c70b84903c80ad8fa330db35","title":"継続事前学習","pathname":"/docs/jp/ji-ben/continued-pretraining","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"267b","description":"継続ファインチューニングとも呼ばれます。Unslothを使うと継続的に事前学習でき、モデルが新しい言語を学習できます。","breadcrumbs":[{"label":"基本"}]},{"id":"a6c23d77a33eed11aa325cf186374365fbac069c","title":"最後のチェックポイントからのファインチューニング","pathname":"/docs/jp/ji-ben/finetuning-from-last-checkpoint","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f3c1","description":"チェックポイント保存により、ファインチューニングの進捗を保存して、一時停止してから続行できます。","breadcrumbs":[{"label":"基本"}]},{"id":"1a67dc318c6e7c970b0bed83a61a7f043ca05676","title":"Unslothベンチマーク","pathname":"/docs/jp/ji-ben/unsloth-benchmarks","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"1f4ca","description":"NVIDIA GPU上で記録されたUnslothのベンチマーク。","breadcrumbs":[{"label":"基本"}]},{"id":"3864bcb37ba6a47277fdce43a7f1d4bc977d7edb","title":"OpenCodeでローカルAIモデルを実行する方法","pathname":"/docs/jp/tong-he/opencode","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"rectangle-vertical","description":"ローカルデバイスでOpenCodeとオープンLLMを接続するためのガイド。","breadcrumbs":[{"label":"統合"}]},{"id":"1040e353b555381fe4f250e6417f1e51602685b2","title":"OpenClawでローカルAIモデルを実行する方法","pathname":"/docs/jp/tong-he/openclaw","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"lobster","description":"OpenClawを使ってローカルLLMを実行するためのガイド。","breadcrumbs":[{"label":"統合"}]},{"id":"a624ad2fe3b378ac0bbfee74cfdda2550bac8b5d","title":"Hermes AgentでローカルAIモデルを実行する方法","pathname":"/docs/jp/tong-he/hermes-agent","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"caduceus","description":"Hermes AgentでオープンLLMをローカルで使うためのガイド。","breadcrumbs":[{"label":"統合"}]},{"id":"1cb063d078a6ee398eff3b5e436d43e1fb2d6e52","title":"Python SDKをUnslothに接続","pathname":"/docs/jp/tong-he/python-sdkwounslothni","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"python","description":"ストリーミング、ビジョン、関数呼び出し、Unslothのサーバーサイド組み込みツールを含む、公式OpenAIまたはAnthropic SDKを使ってPythonからUnslothのローカルAPIを呼び出すためのガイド。","breadcrumbs":[{"label":"統合"}]},{"id":"f70478a97a94b1f315a2c502fe67fb5f4f746cb9","title":"CurlとHTTPをUnslothに接続","pathname":"/docs/jp/tong-he/curltohttpwounslothni","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"spiral","description":"curl（または任意のHTTPクライアント）でUnslothのAPIにアクセスするためのガイド。すべてのエンドポイントと機能について、コピペ可能なレシピ付きです。","breadcrumbs":[{"label":"統合"}]},{"id":"b1418ee8202cf9a87ebe6dc21e8d3a6900e24405","title":"Unsloth Kernels + PackingでLLM学習を3倍高速化","pathname":"/docs/jp/burogu/3x-faster-training-packing","siteSpaceId":"sitesp_8AL84","lang":"ja","emoji":"26a1","description":"Unslothが学習スループットを向上させ、ファインチューニング時のパディング無駄をなくす方法を学びましょう。","breadcrumbs":[{"label":"ブログ"}]},{"id":"243043ca15b2f3f3a09863fbaadc89c1b0b6720b","title":"50万コンテキスト長のファインチューニング","pathname":"/docs/jp/burogu/500k-context-length-fine-tuning","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"ruler-combined","description":"Unslothで50万以上のトークンのコンテキストウィンドウを有効にしてファインチューニングする方法を学びましょう。","breadcrumbs":[{"label":"ブログ"}]},{"id":"eb4e6f009b30d4f300c074f6d5886aac9d47065a","title":"量子化対応学習（QAT）","pathname":"/docs/jp/burogu/quantization-aware-training-qat","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"down-left-and-up-right-to-center","description":"精度を回復するために、UnslothとPyTorchでモデルを4-bitに量子化します。","breadcrumbs":[{"label":"ブログ"}]},{"id":"23f137d60bd35d4f00b21fe55329fe6e6909b293","title":"NVIDIA DGX StationでUnslothを使ってLLMをファインチューニング","pathname":"/docs/jp/burogu/dgx-station","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"microchip-ai","description":"Unslothのノートブックを使ってファインチューニングする方法のNVIDIA DGX Stationチュートリアル。","breadcrumbs":[{"label":"ブログ"}]},{"id":"ab3540db05c75f06746429e2d7defafc52bf32e1","title":"UnslothとDockerでLLMをファインチューニングする方法","pathname":"/docs/jp/burogu/how-to-fine-tune-llms-with-unsloth-and-docker","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"docker","description":"UnslothのDockerイメージを使ってLLMをファインチューニングするか、強化学習（RL）を行う方法を学びましょう。","breadcrumbs":[{"label":"ブログ"}]},{"id":"1df4a60634b989f58e2dabb3b331e7cca93c8e5b","title":"NVIDIA DGX SparkとUnslothでLLMをファインチューニング","pathname":"/docs/jp/burogu/fine-tuning-llms-with-nvidia-dgx-spark-and-unsloth","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"sparkle","description":"NVIDIA DGX Spark上でOpenAI gpt-ossを使ってファインチューニングと強化学習（RL）を行うチュートリアル。","breadcrumbs":[{"label":"ブログ"}]},{"id":"2356c4886c1153e7e0c3288e404d25e96bc02144","title":"Blackwell、RTX 50シリーズとUnslothでLLMをファインチューニング","pathname":"/docs/jp/burogu/fine-tuning-llms-with-blackwell-rtx-50-series-and-unsloth","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"microchip","description":"NVIDIAのBlackwell RTX 50シリーズとB200 GPUでLLMをファインチューニングする方法を、ステップバイステップガイドで学びましょう。","breadcrumbs":[{"label":"ブログ"}]},{"id":"4bc9c33936093721ed847dcfd65023a68edf2530","title":"AMDの力を解き放つ: Unslothの公式サポートが登場！","pathname":"/docs/jp/burogu/amdnowokitsu-unslothnosaptoga","siteSpaceId":"sitesp_8AL84","lang":"ja","icon":"square-up-right","description":"UnslothのAMD GPUサポートが正式対応になりました。NVIDIAハードウェア不要で、約70%少ないメモリで最大2倍高速にLLMをファインチューニングできます。","breadcrumbs":[{"label":"ブログ"}]},{"id":"9eee90080b64494b2b4f2d69662dc2dbef8c74c8","title":"Unsloth-Dokumentation","pathname":"/docs/de","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9a5","description":"Unsloth ist ein Open-Source-Framework zum Ausführen und Trainieren von Modellen.","breadcrumbs":[{"label":"Loslegen"}]},{"id":"2e5c28b5681fb8721dd9abf2ead7595b01aea335","title":"Feinabstimmung für Anfänger","pathname":"/docs/de/loslegen/fine-tuning-for-beginners","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"2b50","description":"","breadcrumbs":[{"label":"Loslegen"}]},{"id":"53df1382354c7ace1c37120c4af6ed50511854dd","title":"Unsloth-Anforderungen","pathname":"/docs/de/loslegen/fine-tuning-for-beginners/unsloth-requirements","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f6e0","description":"Hier sind die Anforderungen von Unsloth, einschließlich System- und GPU-VRAM-Anforderungen.","breadcrumbs":[{"label":"Loslegen"},{"label":"Feinabstimmung für Anfänger","emoji":"2b50"}]},{"id":"247e762fc931f96d3998ecfa1a4402cf524e9e97","title":"FAQ + Ist Feinabstimmung das Richtige für mich?","pathname":"/docs/de/loslegen/fine-tuning-for-beginners/faq-+-is-fine-tuning-right-for-me","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f914","description":"Wenn du unsicher bist, ob Feinabstimmung das Richtige für dich ist, sieh hier nach! Erfahre mehr über Missverständnisse zur Feinabstimmung, wie sie sich im Vergleich zu RAG verhält und mehr:","breadcrumbs":[{"label":"Loslegen"},{"label":"Feinabstimmung für Anfänger","emoji":"2b50"}]},{"id":"c96e3433e67c1b26226b1118128145a6ff8a990a","title":"Unsloth-Notebooks","pathname":"/docs/de/loslegen/unsloth-notebooks","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4d2","description":"Notebooks zur Feinabstimmung: Entdecke den Unsloth-Katalog.","breadcrumbs":[{"label":"Loslegen"}]},{"id":"0d4e311c01cc0577c64d15b8a41f22ba29eab7fd","title":"Unsloth-Modellkatalog","pathname":"/docs/de/loslegen/unsloth-model-catalog","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f52e","description":"","breadcrumbs":[{"label":"Loslegen"}]},{"id":"fd362a47f0e8cc55190dad52421d6fe66df3a5cb","title":"Unsloth-Installation","pathname":"/docs/de/loslegen/install","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4e5","description":"Lerne, Unsloth lokal oder online zu installieren.","breadcrumbs":[{"label":"Loslegen"}]},{"id":"f2e9c46bf9628a73605243a4f5d2859cf6831957","title":"Unsloth via pip und uv installieren","pathname":"/docs/de/loslegen/install/pip-install","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"desktop-arrow-down","description":"Um Unsloth lokal über Pip zu installieren, befolge die folgenden Schritte:","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"c0a9741cebd0f6061894641559ed770f1819d0ae","title":"Unsloth auf MacOS installieren","pathname":"/docs/de/loslegen/install/mac","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"apple","description":"","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"44f4bf2eea97ad3270d9e11ab9d33f2b9479ae22","title":"Wie man LLMs unter Windows mit Unsloth feinabstimmt (Schritt-für-Schritt-Anleitung)","pathname":"/docs/de/loslegen/install/windows-installation","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"windows","description":"Sieh, wie du Unsloth unter Windows installierst, um lokal mit der Feinabstimmung von LLMs zu beginnen.","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"4ce2c4dc8c5cacd60164238ecd513f358a061597","title":"Unsloth via Docker installieren","pathname":"/docs/de/loslegen/install/docker","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"docker","description":"Unsloth mit unserem offiziellen Docker-Container installieren","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"5216ff7aa081a61171fc75301bb42be87b4a9c8a","title":"Unsloth aktualisieren","pathname":"/docs/de/loslegen/install/updating","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"arrow-rotate-right","description":"Um eine alte Version von Unsloth zu aktualisieren oder zu verwenden, befolge die folgenden Schritte:","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"6cfc4e7277de7763f13423fb2a9191134f5aba98","title":"Leitfaden zur Feinabstimmung von LLMs auf AMD-GPUs mit Unsloth","pathname":"/docs/de/loslegen/install/amd","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"square-up-right","description":"Lerne, wie man große Sprachmodelle (LLMs) auf AMD-GPUs mit Unsloth feinabstimmt.","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"ae33a3cb262bdf9cbd68b7b4495db5c0a6bb5342","title":"LLMs auf Intel-GPUs mit Unsloth feinabstimmen","pathname":"/docs/de/loslegen/install/intel","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"info","description":"Lerne, wie man große Sprachmodelle auf Intel-GPUs trainiert und feinabstimmt.","breadcrumbs":[{"label":"Loslegen"},{"label":"Unsloth-Installation","emoji":"1f4e5"}]},{"id":"44aed34263310d67280841ab6b72ea1e5648761f","title":"Leitfaden zur Feinabstimmung von LLMs","pathname":"/docs/de/loslegen/fine-tuning-llms-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9ec","description":"Lerne alle Grundlagen und Best Practices der Feinabstimmung. Anfängerfreundlich.","breadcrumbs":[{"label":"Loslegen"}]},{"id":"079b6ad30e8c25b4a6caae0d2dc5378a166d54c9","title":"Leitfaden zu Datensätzen","pathname":"/docs/de/loslegen/fine-tuning-llms-guide/datasets-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4c8","description":"Lerne, wie man einen Datensatz für die Feinabstimmung erstellt und vorbereitet.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zur Feinabstimmung von LLMs","emoji":"1f9ec"}]},{"id":"ce825bbf83c91ef73a7fc71d696bd3d1ecc78590","title":"Leitfaden zu Hyperparametern für LoRA-Feinabstimmung","pathname":"/docs/de/loslegen/fine-tuning-llms-guide/lora-hyperparameters-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9e0","description":"Lerne Schritt für Schritt die besten Einstellungen für die Feinabstimmung von LLMs – LoRA-Rang & Alpha, Epochen, Batchgröße + Gradient Accumulation, QLoRA vs. LoRA, Zielmodule und mehr.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zur Feinabstimmung von LLMs","emoji":"1f9ec"}]},{"id":"e74e48e68b0d725224ca82d9e12183f70ec3dd62","title":"Welches Modell sollte ich für die Feinabstimmung verwenden?","pathname":"/docs/de/loslegen/fine-tuning-llms-guide/what-model-should-i-use","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"2753","description":"","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zur Feinabstimmung von LLMs","emoji":"1f9ec"}]},{"id":"3c1de2711fe52945f90b888765c1914adeb1e704","title":"Tutorial: Wie man Llama-3 feinabstimmt und in Ollama verwendet","pathname":"/docs/de/loslegen/fine-tuning-llms-guide/tutorial-how-to-finetune-llama-3-and-use-in-ollama","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f999","description":"Anfängerleitfaden zum Erstellen eines maßgeschneiderten persönlichen Assistenten (wie ChatGPT), der lokal in Ollama läuft","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zur Feinabstimmung von LLMs","emoji":"1f9ec"}]},{"id":"5b0f8932321e8767d629a7ca0f24c3e9add748f5","title":"Leitfaden zu Reinforcement Learning (RL)","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4a1","description":"Lerne alles über Reinforcement Learning (RL) und wie du mit Unsloth und GRPO dein eigenes DeepSeek-R1-Reasoning-Modell trainierst. Ein vollständiger Leitfaden von Anfänger bis Fortgeschrittene.","breadcrumbs":[{"label":"Loslegen"}]},{"id":"c040fd1976b219674df870b5332c0ed4f1ad488c","title":"Reinforcement Learning GRPO mit 7x längerem Kontext","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/grpo-long-context","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f300","description":"Lerne, wie Unsloth ultralange Kontext-RL-Feinabstimmung ermöglicht.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"b6b5ca32e463fa09c0f26919da403e22381129bc","title":"Vision Reinforcement Learning (VLM RL)","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/vision-reinforcement-learning-vlm-rl","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f441-1f5e8","description":"Trainiere Vision-/Multimodal-Modelle via GRPO und RL mit Unsloth!","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"e8e92128cfff3357409da6d50baeeb3180e259ec","title":"FP8 Reinforcement Learning","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/fp8-reinforcement-learning","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f3b1","description":"Trainiere Reinforcement Learning (RL) und GRPO mit FP8-Präzision mit Unsloth.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"2b7904fd5422b954809fd2460ec92281ba56a6fa","title":"Tutorial: Trainiere dein eigenes Reasoning-Modell mit GRPO","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/tutorial-train-your-own-reasoning-model-with-grpo","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"26a1","description":"Anfängerleitfaden, um ein Modell wie Llama 3.1 (8B) mithilfe von Unsloth und GRPO in ein Reasoning-Modell zu verwandeln.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"5b769b691c2ac2681f88fb5053c526306a2c1629","title":"Dokumentation für fortgeschrittenes Reinforcement Learning","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/advanced-rl-documentation","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9e9","description":"Erweiterte Dokumentationseinstellungen bei der Verwendung von Unsloth mit GRPO.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"e2d862d499f3c8adf52168e9caf20c41478a70da","title":"GSPO Reinforcement Learning","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/advanced-rl-documentation/gspo-reinforcement-learning","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"lightbulb-on","description":"Trainiere mit GSPO (Group Sequence Policy Optimization) RL in Unsloth.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"},{"label":"Dokumentation für fortgeschrittenes Reinforcement Learning","emoji":"1f9e9"}]},{"id":"547730be6638817f57e57c3a4e83ced021c3bd5b","title":"RL Reward Hacking","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/advanced-rl-documentation/rl-reward-hacking","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"treasure-chest","description":"Lerne, was Reward Hacking im Reinforcement Learning ist und wie man es bekämpft.","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"},{"label":"Dokumentation für fortgeschrittenes Reinforcement Learning","emoji":"1f9e9"}]},{"id":"d1438bae8d51e95e98d98855d18199f86e47ffa1","title":"FP16 vs BF16 für RL","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/advanced-rl-documentation/fp16-vs-bf16-for-rl","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"2049","description":"Defeating the Training-Inference Mismatch via FP16 https://arxiv.org/pdf/2510.26788 zeigt, dass die Verwendung von float16 besser ist als bfloat16","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"},{"label":"Dokumentation für fortgeschrittenes Reinforcement Learning","emoji":"1f9e9"}]},{"id":"8ad88c458cbd2d3022808d6ae9392f3d311d2c81","title":"Speichereffizientes RL","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/memory-efficient-rl","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"memory","description":"","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"43fcd2dba1e05f87b102b4cfb7f0994eeff8a1db","title":"Training zur Präferenzoptimierung - DPO, ORPO & KTO","pathname":"/docs/de/loslegen/reinforcement-learning-rl-guide/preference-dpo-orpo-and-kto","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f3c6","description":"Lerne über Präferenzabstimmungs-Feinabstimmung mit DPO, GRPO, ORPO oder KTO über Unsloth; befolge die folgenden Schritte:","breadcrumbs":[{"label":"Loslegen"},{"label":"Leitfaden zu Reinforcement Learning (RL)","emoji":"1f4a1"}]},{"id":"0eb7cfce0eb5651e720ac0944c9e36f7124bc8de","title":"Unsloth Studio vorstellen","pathname":"/docs/de/neu/studio","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9a5","description":"KI-Modelle lokal mit Unsloth Studio ausführen und trainieren.","breadcrumbs":[{"label":"Neu"}]},{"id":"bf6b637872d3a9f534daa54866ec7d33d9b369a2","title":"Erste Schritte mit Unsloth Studio","pathname":"/docs/de/neu/studio/start","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"bolt","description":"Ein Leitfaden für den Einstieg in das Feinabstimmungs-Studio, Datenrezepte, den Modularexport und Chat.","breadcrumbs":[{"label":"Neu"},{"label":"Unsloth Studio vorstellen","emoji":"1f9a5"}]},{"id":"73168493b14a89fa27e95b776ade6bd93679a3e5","title":"Wie man Modelle mit Unsloth Studio ausführt","pathname":"/docs/de/neu/studio/chat","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"comment-dots","description":"KI-Modelle, LLMs und GGUFs lokal mit Unsloth Studio ausführen.","breadcrumbs":[{"label":"Neu"},{"label":"Unsloth Studio vorstellen","emoji":"1f9a5"}]},{"id":"fd472c5a40e73b8e54cee17244010f121bf38286","title":"Unsloth Studio Installation","pathname":"/docs/de/neu/studio/install","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"arrow-down-to-square","description":"Lerne, wie man Unsloth Studio auf deinem lokalen Gerät installiert.","breadcrumbs":[{"label":"Neu"},{"label":"Unsloth Studio vorstellen","emoji":"1f9a5"}]},{"id":"dd9f9c69fff351709e4886e820d7d6facf3ec2b3","title":"Unsloth-Datenrezepte","pathname":"/docs/de/neu/studio/data-recipe","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"hat-chef","description":"Lerne, wie man mit den Datenrezepten von Unsloth Studio Datensätze erstellt, aufbaut und bearbeitet.","breadcrumbs":[{"label":"Neu"},{"label":"Unsloth Studio vorstellen","emoji":"1f9a5"}]},{"id":"13c9d0063a9732a68734b74792f3e30153873bf4","title":"Modelle mit Unsloth Studio exportieren","pathname":"/docs/de/neu/studio/export","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"box-isometric","description":"Lerne, wie du deine Safetensor- oder LoRA-Modelldateien in GGUF oder andere Formate exportierst.","breadcrumbs":[{"label":"Neu"},{"label":"Unsloth Studio vorstellen","emoji":"1f9a5"}]},{"id":"baf914c5df639b5b868baac27a080b70396d87c2","title":"Unsloth-Updates","pathname":"/docs/de/neu/changelog","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"sparkles","description":"Unsloth-Changelog für unsere neuesten Releases, Verbesserungen und Fehlerbehebungen.","breadcrumbs":[{"label":"Neu"}]},{"id":"efc00d6b1d286a029d0eec8a5a6a24d50b063840","title":"Qwen3.6 - Wie man lokal ausführt","pathname":"/docs/de/modelle/qwen3.6","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f49c","description":"Führe die neuen Modelle Qwen3.6-27B und 35B-A3B lokal aus!","breadcrumbs":[{"label":"Modelle"}]},{"id":"774ab7ebd11d30a8067d492668e9dd61a8b209fb","title":"Gemma 4 - Wie man lokal ausführt","pathname":"/docs/de/modelle/gemma-4","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"2728","description":"Führe Googles neue Gemma-4-Modelle lokal aus, einschließlich E2B, E4B, 26B A4B und 31B.","breadcrumbs":[{"label":"Modelle"}]},{"id":"6b0583a8449166f2d11877ed16e0b6c150d61586","title":"Gemma 4 Leitfaden zur Feinabstimmung","pathname":"/docs/de/modelle/gemma-4/train","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"flask-gear","description":"Trainiere Gemma 4 von Google mit Unsloth.","breadcrumbs":[{"label":"Modelle"},{"label":"Gemma 4 - Wie man lokal ausführt","emoji":"2728"}]},{"id":"78057bfde350b3d5e5b61e20e35b56d077a87d6e","title":"NVIDIA Nemotron 3 Nano Omni - Wie man lokal ausführt","pathname":"/docs/de/modelle/nemotron-3-nano-omni","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9e9","description":"Führe und feinabstimme Nemotron-3-Nano-Omni-30B-A3B lokal auf deinem Gerät!","breadcrumbs":[{"label":"Modelle"}]},{"id":"66ff370179820adad5a2dd2711d47bd05d4bbe2e","title":"Kimi K2.6 - Wie man lokal ausführt","pathname":"/docs/de/modelle/kimi-k2.6","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f95d","description":"Schritt-für-Schritt-Anleitung zum Ausführen von Kimi-K2.6 auf deinem eigenen lokalen Gerät.","breadcrumbs":[{"label":"Modelle"}]},{"id":"53f3f20fc7a21f1ca51eb7268267793cb5975b35","title":"Qwen3.5 - Wie man lokal ausführt","pathname":"/docs/de/modelle/qwen3.5","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f49c","description":"Führe die neuen Qwen3.5-LLMs aus, darunter Medium: Qwen3.5-35B-A3B, 27B, 122B-A10B, Small: Qwen3.5-0.8B, 2B, 4B, 9B und 397B-A17B auf deinem lokalen Gerät!","breadcrumbs":[{"label":"Modelle"}]},{"id":"5e52a979e20c6eaea490f46fcf09c99fd97ae449","title":"Qwen3.5 Leitfaden zur Feinabstimmung","pathname":"/docs/de/modelle/qwen3.5/fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"flask-gear","description":"Lerne, wie man Qwen3.5-LLMs mit Unsloth feinabstimmt.","breadcrumbs":[{"label":"Modelle"},{"label":"Qwen3.5 - Wie man lokal ausführt","emoji":"1f49c"}]},{"id":"fffcdfcc3dce216f018aa6cc7d0b3de63f9dd887","title":"Qwen3.5 GGUF-Benchmarks","pathname":"/docs/de/modelle/qwen3.5/gguf-benchmarks","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"chart-fft","description":"Sieh, wie sich die dynamischen GGUFs von Unsloth schlagen + Analyse von Perplexity, KL-Divergenz und MXFP4.","breadcrumbs":[{"label":"Modelle"},{"label":"Qwen3.5 - Wie man lokal ausführt","emoji":"1f49c"}]},{"id":"3798c366d7b3cb260558eaaf89e704289df7e2fc","title":"GLM-5.1 - Wie man lokal ausführt","pathname":"/docs/de/modelle/glm-5.1","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"z","description":"Führe das neue GLM-5.1-Modell von Z.ai auf deinem eigenen lokalen Gerät aus!","breadcrumbs":[{"label":"Modelle"}]},{"id":"9b29615ab338f1d1924174468718cfeab406f641","title":"Tutorials zu großen Sprachmodellen (LLMs)","pathname":"/docs/de/modelle/tutorials","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f680","description":"","breadcrumbs":[{"label":"Modelle"}]},{"id":"577dc1f2e661d959da353a9d4b3748912873d42e","title":"Qwen3 - Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/qwen3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f320","description":"Lerne, Qwen3 lokal mit Unsloth + unseren Dynamic 2.0 Quants auszuführen und feinabzustimmen","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"e81b513117b7976d1bb2a959ddaaaf797b39b9e0","title":"Qwen3-VL: Leitfaden zum Ausführen","pathname":"/docs/de/modelle/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-vl-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f320","description":"Lerne, Qwen3-VL lokal mit Unsloth feinabzustimmen und auszuführen.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"},{"label":"Qwen3 - Wie man ausführt und feinabstimmt","emoji":"1f320"}]},{"id":"a1f4ef520ffb1a185e8a6deff4ec2b5569d8f5a7","title":"Qwen3-2507: Lokaler Ausführungsleitfaden","pathname":"/docs/de/modelle/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-2507","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f320","description":"Führe Qwen3-30B-A3B-2507 und die 235B-A22B Thinking- und Instruct-Versionen lokal auf deinem Gerät aus!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"},{"label":"Qwen3 - Wie man ausführt und feinabstimmt","emoji":"1f320"}]},{"id":"f778dfc3ae4955798c6678a508ac2c8b21a949a5","title":"MiniMax-M2.7 - Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/minimax-m27","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"waveform","description":"Führe das MiniMax-M2.7-LLM lokal auf deinem eigenen Gerät aus!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"f9b4b78ffbc074c7f10ffb0ff9b54f66e629a834","title":"GLM-5: Leitfaden zum Ausführen lokal","pathname":"/docs/de/modelle/tutorials/glm-5","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"z","description":"Führe das neue GLM-5-Modell von Z.ai auf deinem eigenen lokalen Gerät aus!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"f26cd5b2ebe566940595d2276192dd0ced0b1b5e","title":"Kimi K2.5: Leitfaden zum Ausführen lokal","pathname":"/docs/de/modelle/tutorials/kimi-k2.5","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f95d","description":"Leitfaden zum Ausführen von Kimi-K2.5 auf deinem eigenen lokalen Gerät!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"1e435114456bba34bc232a642cd33933aee06bbc","title":"GLM-4.7-Flash: Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/glm-4.7-flash","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"z","description":"Führe GLM-4.7-Flash lokal auf deinem Gerät aus und feinabstimme es!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"db6f61dbf018d9aa3e178cc7e880c0130989bce0","title":"MiniMax-M2.5: Leitfaden zum Ausführen","pathname":"/docs/de/modelle/tutorials/minimax-m25","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"waveform","description":"Führe MiniMax-M2.5 lokal auf deinem eigenen Gerät aus!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"542ecfd3ee96a444791bde7f395bdd1a59ec1ea5","title":"Qwen3-Coder: Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/qwen3-coder-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f320","description":"Führe Qwen3-Coder-30B-A3B-Instruct und 480B-A35B lokal mit den dynamischen Quants von Unsloth aus.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"ed8345f0cc9e0e67f0b2d0992b6381ecd6a9899b","title":"Gemma 3 - Leitfaden zum Ausführen","pathname":"/docs/de/modelle/tutorials/gemma-3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"google","description":"Wie man Gemma 3 effektiv mit unseren GGUFs in llama.cpp, Ollama, Open WebUI ausführt und wie man mit Unsloth feinabstimmt!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"6375cb14400f983c1ec3d3c0d5f8699e3ee9a5ae","title":"Gemma 3n: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/gemma-3-how-to-run-and-fine-tune/gemma-3n-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"google","description":"Führe Googles neues Gemma 3n lokal mit Dynamic GGUFs in llama.cpp, Ollama, Open WebUI aus und feinabstimme es mit Unsloth!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"},{"label":"Gemma 3 - Leitfaden zum Ausführen","icon":"google"}]},{"id":"1e7861e594f2ea678a28332062a23bcee10c4e3d","title":"DeepSeek-OCR 2: Leitfaden zum Ausführen und Feinabstimmen","pathname":"/docs/de/modelle/tutorials/deepseek-ocr-2","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f433","description":"Leitfaden zum Ausführen und Feinabstimmen von DeepSeek-OCR-2 lokal.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"9eaa51932747df2c1f3c4a4eb10410fa2cb5a2dd","title":"GLM-4.7: Leitfaden zum Ausführen lokal","pathname":"/docs/de/modelle/tutorials/glm-4.7","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"z","description":"Ein Leitfaden, wie man das Z.ai-GLM-4.7-Modell auf deinem eigenen lokalen Gerät ausführt!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"f9a59ced7d21e1bde6109eb7cfa0f721e1602428","title":"Wie man Qwen-Image-2512 lokal in ComfyUI ausführt","pathname":"/docs/de/modelle/tutorials/qwen-image-2512","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f49f","description":"Schritt-für-Schritt-Tutorial zum Ausführen von Qwen-Image-2512 auf deinem lokalen Gerät mit ComfyUI.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"34e394724db99601f06f4a6d709b709fc4f49ab5","title":"Qwen-Image-2512 in stable-diffusion.cpp ausführen Tutorial","pathname":"/docs/de/modelle/tutorials/qwen-image-2512/stable-diffusion.cpp","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f3a8","description":"Tutorial zur Verwendung von Qwen-Image-2512 in stable-diffusion.cpp.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"},{"label":"Wie man Qwen-Image-2512 lokal in ComfyUI ausführt","emoji":"1f49f"}]},{"id":"f092a1211912dae2208951f680a2b5a9944a6af4","title":"Devstral 2 - Leitfaden zum Ausführen","pathname":"/docs/de/modelle/tutorials/devstral-2","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4d9","description":"Leitfaden zum lokalen Ausführen der Mistral-Devstral-2-Modelle: 123B-Instruct-2512 und Small-2-24B-Instruct-2512.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"ce1a52c32558234afe580939db0fcfe220dc208b","title":"Ministral 3 - Leitfaden zum Ausführen","pathname":"/docs/de/modelle/tutorials/ministral-3","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f431","description":"Leitfaden für Mistral-Ministral-3-Modelle, um sie lokal auf deinem Gerät auszuführen oder feinabzustimmen","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"ba9915a2be20805eb5ba7d6ebec553163ddf863e","title":"DeepSeek-OCR: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/deepseek-ocr-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f433","description":"Leitfaden zum Ausführen und Feinabstimmen von DeepSeek-OCR lokal.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"ec4485f0cffa8c16963f28544e89a652f988d869","title":"Kimi K2 Thinking: Leitfaden zum lokalen Ausführen","pathname":"/docs/de/modelle/tutorials/kimi-k2-thinking-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f319","description":"Leitfaden zum Ausführen von Kimi-K2-Thinking und Kimi-K2 auf deinem eigenen lokalen Gerät!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"d1a24e17aacb2e1edef96e954b901fe1f785221e","title":"GLM-4.6: Leitfaden zum lokalen Ausführen","pathname":"/docs/de/modelle/tutorials/glm-4.6-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"z","description":"Ein Leitfaden, wie man das Z.ai-GLM-4.6- und GLM-4.6V-Flash-Modell auf deinem eigenen lokalen Gerät ausführt!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"c8b6895be773f2849bd57d1b0d3578f491bfa67c","title":"Qwen3-Next: Leitfaden zum lokalen Ausführen","pathname":"/docs/de/modelle/tutorials/qwen3-next","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f320","description":"Führe Qwen3-Next-80B-A3B-Instruct- und Thinking-Versionen lokal auf deinem Gerät aus!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"8ecff6350d3f4cbfb3de937da82a372c1385dcc6","title":"FunctionGemma: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/functiongemma","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"google","description":"Lerne, wie man FunctionGemma lokal auf deinem Gerät und Telefon ausführt und feinabstimmt.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"42aa6b132b83bbca34de4b7a6d3d2074272827c3","title":"DeepSeek-V3.1: Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/deepseek-v3.1-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f40b","description":"Ein Leitfaden, wie man DeepSeek-V3.1 und Terminus auf deinem eigenen lokalen Gerät ausführt!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"2d93c1e9c4de5c5a267893947e9721f008ed9418","title":"DeepSeek-R1-0528: Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/deepseek-r1-0528-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f40b","description":"Ein Leitfaden, wie man DeepSeek-R1-0528 einschließlich Qwen3 auf deinem eigenen lokalen Gerät ausführt!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"0ba99f57da4d961b4b6ac0ebc82f8fc360f57cc1","title":"Liquid LFM2.5: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/lfm2.5","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4a7","description":"Führe LFM2.5 Instruct und Vision lokal auf deinem Gerät aus und feinabstimme sie!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"a98dac686894198901004645b13c5c4c51f32843","title":"Magistral: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/magistral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4a5","description":"Lerne Magistral kennen – Mistrals neue Reasoning-Modelle.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"ed18e39c4ad9797b97e59c64acb1e3bee4b53103","title":"IBM Granite 4.0","pathname":"/docs/de/modelle/tutorials/ibm-granite-4.0","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"cube","description":"Wie man IBM Granite-4.0 mit Unsloth-GGUFs in llama.cpp und Ollama ausführt und wie man feinabstimmt!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"48def9956d09f2f427c4285d05bd7567ca706ce8","title":"Llama 4: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/llama-4-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f999","description":"Wie man Llama 4 lokal mit unseren dynamischen GGUFs ausführt, die im Vergleich zur Standard-Quantisierung die Genauigkeit verbessern.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"4d0909d6bdc8700fe155476c8a8804eb7bcb39a7","title":"Grok 2","pathname":"/docs/de/modelle/tutorials/grok-2","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"square-x-twitter","description":"Führe xAIs Grok-2-Modell lokal aus!","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"7f1818c639a40852ccefc40e5bc9f330d43c0920","title":"Devstral: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/devstral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4d9","description":"Führe Mistral Devstral 1.1 aus und feinabstimme es, einschließlich Small-2507 und 2505.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"69b29df1a649bc85fbd042bb58c2d5bc609ad71b","title":"Wie man lokale LLMs mit Docker ausführt: Schritt-für-Schritt-Anleitung","pathname":"/docs/de/modelle/tutorials/how-to-run-llms-with-docker","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"docker","description":"Lerne, wie man Large Language Models (LLMs) mit Docker und Unsloth auf deinem lokalen Gerät ausführt.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"6874079159971d44a046544779789075f4168545","title":"DeepSeek-V3-0324: Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/deepseek-v3-0324-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f433","description":"Wie man DeepSeek-V3-0324 lokal mit unseren dynamischen Quants ausführt, die die Genauigkeit verbessern","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"59bfd1b66766603cea7f531c9b7fd8669737f1a6","title":"DeepSeek-R1: Wie man lokal ausführt","pathname":"/docs/de/modelle/tutorials/deepseek-r1-how-to-run-locally","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f40b","description":"Ein Leitfaden, wie du unsere 1,58-Bit-Dynamic-Quants für DeepSeek-R1 mit llama.cpp ausführen kannst.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"6996e8b117edf77aa284fea26ffc8d3c9f4a6076","title":"DeepSeek-R1 Dynamisch 1,58-Bit","pathname":"/docs/de/modelle/tutorials/deepseek-r1-how-to-run-locally/deepseek-r1-dynamic-1.58-bit","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f433","description":"Sieh Vergleichstabellen zur Leistung von Unsloths dynamischen GGUF-Quants gegenüber standardmäßigen IMatrix-Quants.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"},{"label":"DeepSeek-R1: Wie man lokal ausführt","emoji":"1f40b"}]},{"id":"9f9221053a31dd3d971a43c6bc3f7afc4b2ca60a","title":"Phi-4 Reasoning: Wie man ausführt und feinabstimmt","pathname":"/docs/de/modelle/tutorials/phi-4-reasoning-how-to-run-and-fine-tune","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"windows","description":"Lerne, Phi-4-Reasoning-Modelle lokal mit Unsloth + unseren Dynamic 2.0 Quants auszuführen und feinabzustimmen","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"57112ad35933fe550381d802d8f1e47ff8533916","title":"QwQ-32B: Wie man effektiv ausführt","pathname":"/docs/de/modelle/tutorials/qwq-32b-how-to-run-effectively","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f320","description":"Wie man QwQ-32B effektiv mit unseren Fehlerbehebungen und ohne endlose Generierungen + GGUFs ausführt.","breadcrumbs":[{"label":"Modelle"},{"label":"Tutorials zu großen Sprachmodellen (LLMs)","emoji":"1f680"}]},{"id":"70002e65495a1a44f155c9dabf1fb52b9d84de66","title":"Wie man Unsloth als API-Endpunkt verwendet","pathname":"/docs/de/grundlagen/api","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"globe","description":"","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"03532de69dfe0230fe5114e809721d8b7dd74ca6","title":"Inferenz & Bereitstellung","pathname":"/docs/de/grundlagen/inference-and-deployment","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f5a5","description":"Lerne, wie du dein feinabgestimmtes Modell speicherst, damit du es in deiner bevorzugten Inferenz-Engine ausführen kannst.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"9cfeafb2cc359999e3a7f6ba6ffa5468e4752653","title":"In GGUF speichern","pathname":"/docs/de/grundlagen/inference-and-deployment/saving-to-gguf","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"a241be61d5411b52c7d9ab17486a0185949606b1","title":"Spekulatives Decoding","pathname":"/docs/de/grundlagen/inference-and-deployment/saving-to-gguf/speculative-decoding","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Spekulatives Decoding mit llama-server, llama.cpp, vLLM und mehr für 2x schnellere Inferenz","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"},{"label":"In GGUF speichern"}]},{"id":"af094159d1c157db0d9afc00bd98b849fcdb8f0c","title":"Leitfaden für vLLM-Bereitstellung & Inferenz","pathname":"/docs/de/grundlagen/inference-and-deployment/vllm-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Leitfaden zum Speichern und Bereitstellen von LLMs in vLLM, um LLMs produktiv auszuliefern","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"6969b7d50e6d4e4a2fd432fda78971db7d4f929a","title":"vLLM-Engine-Argumente","pathname":"/docs/de/grundlagen/inference-and-deployment/vllm-guide/vllm-engine-arguments","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"},{"label":"Leitfaden für vLLM-Bereitstellung & Inferenz"}]},{"id":"7f6b2c78a8d675a1c775391d78fe85d3d0e31b3a","title":"Leitfaden zum Hot-Swapping von LoRA","pathname":"/docs/de/grundlagen/inference-and-deployment/vllm-guide/lora-hot-swapping-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"},{"label":"Leitfaden für vLLM-Bereitstellung & Inferenz"}]},{"id":"535d0cfbde7f5f4826b1f619981b5b10742e2330","title":"In Ollama speichern","pathname":"/docs/de/grundlagen/inference-and-deployment/saving-to-ollama","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"ea71c5c0df369f2b0055399448c044aabed9efba","title":"Modelle in LM Studio bereitstellen","pathname":"/docs/de/grundlagen/inference-and-deployment/lm-studio","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Modelle in GGUF speichern, damit du sie in LM Studio ausführen und bereitstellen kannst","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"d373cb87b7b6b14d5e135802924614c5ccbbb8ec","title":"Wie man die LM Studio CLI im Linux-Terminal installiert","pathname":"/docs/de/grundlagen/inference-and-deployment/lm-studio/how-to-install-lm-studio-cli-in-linux-terminal","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f47e","description":"Installationsanleitung für die LM Studio CLI ohne UI in einer Terminal-Instanz.","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"},{"label":"Modelle in LM Studio bereitstellen"}]},{"id":"7ee5a053c7473687e3fc8557db16d666991941dc","title":"Leitfaden für SGLang-Bereitstellung & Inferenz","pathname":"/docs/de/grundlagen/inference-and-deployment/sglang-guide","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Leitfaden zum Speichern und Bereitstellen von LLMs in SGLang, um LLMs produktiv auszuliefern","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"6f7dcba79c06230757d338618fe62a8da47076da","title":"Leitfaden zur Bereitstellung von llama-server & OpenAI-Endpunkt","pathname":"/docs/de/grundlagen/inference-and-deployment/llama-server-and-openai-endpoint","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Bereitstellung über llama-server mit einem OpenAI-kompatiblen Endpunkt","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"86b6f77ee96e4edb00c776189ff1bf5032a3aaeb","title":"Wie man LLMs auf deinem iOS- oder Android-Handy ausführt und bereitstellt","pathname":"/docs/de/grundlagen/inference-and-deployment/deploy-llms-phone","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4f1","description":"Tutorial zum Feinabstimmen deines eigenen LLMs und zur Bereitstellung auf deinem Android- oder iPhone mit ExecuTorch.","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"14781be2f5de20d42f8771160661ee0e1bf6e874","title":"Fehlerbehebung bei der Inferenz","pathname":"/docs/de/grundlagen/inference-and-deployment/troubleshooting-inference","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Wenn du Probleme beim Ausführen oder Speichern deines Modells hast.","breadcrumbs":[{"label":"Grundlagen"},{"label":"Inferenz & Bereitstellung","emoji":"1f5a5"}]},{"id":"d12c953ceacbd6c3e44f3aa911056928e0488f5b","title":"Wie man lokale LLMs mit Claude Code ausführt","pathname":"/docs/de/grundlagen/claude-code","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"claude","description":"Leitfaden zur Verwendung offener Modelle mit Claude Code auf deinem lokalen Gerät.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"1813c928d883d651dff92062bc0da6e96d06e50a","title":"Wie man lokale LLMs mit OpenAI Codex ausführt","pathname":"/docs/de/grundlagen/codex","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"openai","description":"Verwende offene Modelle mit OpenAI Codex lokal auf deinem Gerät.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"8b1f9407050c14f70efd7cfaa7d1daf4a67ff3e7","title":"Multi-GPU-Feinabstimmung mit Unsloth","pathname":"/docs/de/grundlagen/multi-gpu-training-with-unsloth","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"rectangle-history","description":"Lerne, wie man LLMs auf mehreren GPUs und mit Parallelisierung mit Unsloth feinabstimmt.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"5514b4f06913bfcee7c5b9d6e0fd6c014340cdf3","title":"Multi-GPU-Feinabstimmung mit Distributed Data Parallel (DDP)","pathname":"/docs/de/grundlagen/multi-gpu-training-with-unsloth/ddp","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Lerne, wie man die Unsloth-CLI verwendet, um mit Distributed Data Parallel (DDP) auf mehreren GPUs zu trainieren!","breadcrumbs":[{"label":"Grundlagen"},{"label":"Multi-GPU-Feinabstimmung mit Unsloth","icon":"rectangle-history"}]},{"id":"f9cecc317356c3fa794a39ae4190a6d22c029f46","title":"Leitfaden zur Feinabstimmung von Embedding-Modellen mit Unsloth","pathname":"/docs/de/grundlagen/embedding-finetuning","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f50e","description":"Lerne, wie man Embedding-Modelle einfach mit Unsloth feinabstimmt.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"d2073bad97e9c04578abf5444104a5502ae38941","title":"MoE-Modelle 12x schneller mit Unsloth feinabstimmen","pathname":"/docs/de/grundlagen/faster-moe","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f48e","description":"Trainiere MoE-LLMs lokal mit dem Unsloth-Leitfaden.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"18ba00f910090c7a44c97dedb9a0d82ad0eccacc","title":"Leitfaden zur Feinabstimmung von Text-to-Speech (TTS)","pathname":"/docs/de/grundlagen/text-to-speech-tts-fine-tuning","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f50a","description":"Lerne, wie man TTS- und STT-Sprachmodelle mit Unsloth feinabstimmt.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"abe22dcb3049581e00371a1b82b9e9cf6821a9b0","title":"Unsloth Dynamic 2.0 GGUFs","pathname":"/docs/de/grundlagen/unsloth-dynamic-2.0-ggufs","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9a5","description":"Ein großes neues Upgrade für unsere Dynamic Quants!","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"575776db88d905467eea1b184b6780b2f9ed78e5","title":"Unsloth Dynamic GGUFs auf Aider Polyglot","pathname":"/docs/de/grundlagen/unsloth-dynamic-2.0-ggufs/unsloth-dynamic-ggufs-on-aider-polyglot","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f9a5","description":"Leistung von Unsloth Dynamic GGUFs auf den Aider-Polyglot-Benchmarks","breadcrumbs":[{"label":"Grundlagen"},{"label":"Unsloth Dynamic 2.0 GGUFs","emoji":"1f9a5"}]},{"id":"ba7e51b2382f5cf41d34361579ec54dd6bfc4e71","title":"Leitfaden zur Tool-Aufrufung für lokale LLMs","pathname":"/docs/de/grundlagen/tool-calling-guide-for-local-llms","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"screwdriver-wrench","description":"","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"57170bb26b2747b2ef94d23a7666e8a08d688b27","title":"Vision-Feinabstimmung","pathname":"/docs/de/grundlagen/vision-fine-tuning","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f441","description":"Lerne, wie man Vision-/Multimodal-LLMs mit Unsloth feinabstimmt","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"3fc49911b5923db7fb5a51b50c7c638ebe732f42","title":"Fehlerbehebung & FAQs","pathname":"/docs/de/grundlagen/troubleshooting-and-faqs","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"26a0","description":"Tipps zur Behebung von Problemen und häufig gestellte Fragen.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"e2d19835fc68fb83771756e84c0d98ac1e182677","title":"Hugging Face Hub, XET-Debugging","pathname":"/docs/de/grundlagen/troubleshooting-and-faqs/hugging-face-hub-xet-debugging","siteSpaceId":"sitesp_L6rLB","lang":"de","description":"Debugging, Fehlerbehebung bei hängenden, stecken gebliebenen und langsamen Downloads","breadcrumbs":[{"label":"Grundlagen"},{"label":"Fehlerbehebung & FAQs","emoji":"26a0"}]},{"id":"3757060b6860c23169555a4f9ab57784a90ed455","title":"Chat-Vorlagen","pathname":"/docs/de/grundlagen/chat-templates","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4ac","description":"Lerne die Grundlagen und Anpassungsoptionen von Chat-Vorlagen kennen, einschließlich Conversational-, ChatML-, ShareGPT-, Alpaca-Formate und mehr!","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"4a52add32b0967d961ca39225bc178234e2beea9","title":"Unsloth-Umgebungsflags","pathname":"/docs/de/grundlagen/unsloth-environment-flags","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f6e0","description":"Erweiterte Flags, die nützlich sein könnten, wenn du defekte Feinabstimmungen siehst oder Dinge abschalten möchtest.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"7833952b3c01beb5e61cbcebda8182c1be83777e","title":"Fortgesetztes Vortraining","pathname":"/docs/de/grundlagen/continued-pretraining","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"267b","description":"Auch bekannt als fortgesetzte Feinabstimmung. Unsloth ermöglicht es dir, kontinuierlich vorzutrainen, damit ein Modell eine neue Sprache lernen kann.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"24e353a4cd4d66055901c275613af851f819330c","title":"Feinabstimmung vom letzten Checkpoint","pathname":"/docs/de/grundlagen/finetuning-from-last-checkpoint","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f3c1","description":"Checkpointing erlaubt es dir, deinen Feinabstimmungsfortschritt zu speichern, damit du ihn pausieren und später fortsetzen kannst.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"8f066782276b19e21753a9a2f644fd81a8071c0f","title":"Unsloth-Benchmarks","pathname":"/docs/de/grundlagen/unsloth-benchmarks","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"1f4ca","description":"Von Unsloth aufgezeichnete Benchmarks auf NVIDIA-GPUs.","breadcrumbs":[{"label":"Grundlagen"}]},{"id":"69edb30cadb1e5ecada5bd2a700b63f49e94955f","title":"Wie man lokale KI-Modelle mit OpenCode ausführt","pathname":"/docs/de/integrationen/opencode","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"rectangle-vertical","description":"Leitfaden zum Verbinden offener LLMs mit OpenCode auf deinem lokalen Gerät.","breadcrumbs":[{"label":"Integrationen"}]},{"id":"4ad33e7ce59bb7203a1fddd7e6e0102f70c191db","title":"Wie man lokale KI-Modelle mit OpenClaw ausführt","pathname":"/docs/de/integrationen/openclaw","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"lobster","description":"Leitfaden zum Ausführen lokaler LLMs mit OpenClaw.","breadcrumbs":[{"label":"Integrationen"}]},{"id":"d063f905fd9c48e1a14e204b6afea83a23b390f3","title":"Wie man lokale KI-Modelle mit Hermes Agent ausführt","pathname":"/docs/de/integrationen/hermes-agent","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"caduceus","description":"Leitfaden zur lokalen Verwendung offener LLMs mit Hermes Agent.","breadcrumbs":[{"label":"Integrationen"}]},{"id":"67298920320a528a941c4123bc768cc0a79e7332","title":"Python-SDK mit Unsloth verbinden","pathname":"/docs/de/integrationen/python-sdk-mit-unsloth-verbinden","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"python","description":"Leitfaden zum Aufruf der lokalen API von Unsloth aus Python mit den offiziellen OpenAI- oder Anthropic-SDKs, einschließlich Streaming, Vision, Function Calling und der integrierten serverseitigen Tools von Unsloth.","breadcrumbs":[{"label":"Integrationen"}]},{"id":"cede767fc3b9f7535c2c3cc36963872488a3bd1d","title":"Curl & HTTP mit Unsloth verbinden","pathname":"/docs/de/integrationen/curl-and-http-mit-unsloth-verbinden","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"spiral","description":"Leitfaden, um Unsloths API mit curl (oder jedem HTTP-Client) anzusprechen, komplett mit kopierbaren Rezepten für jeden Endpunkt und jede Funktion.","breadcrumbs":[{"label":"Integrationen"}]},{"id":"c119d057de6a9fb91c2922e5fb7f498aa600bc18","title":"3x schnelleres LLM-Training mit Unsloth-Kernels + Packing","pathname":"/docs/de/blog/3x-faster-training-packing","siteSpaceId":"sitesp_L6rLB","lang":"de","emoji":"26a1","description":"Lerne, wie Unsloth den Trainingsdurchsatz erhöht und Padding-Verschwendung bei der Feinabstimmung beseitigt.","breadcrumbs":[{"label":"Blog"}]},{"id":"d9c1ce434cf33ed7ef7bf701657d32b0e37edee8","title":"500K Kontextlängen-Feinabstimmung","pathname":"/docs/de/blog/500k-context-length-fine-tuning","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"ruler-combined","description":"Lerne, wie man mit Unsloth eine Feinabstimmung mit einem Kontextfenster von über 500K Tokens aktiviert.","breadcrumbs":[{"label":"Blog"}]},{"id":"403e1f81fd936dfabff4d1937d0bbdd84585c969","title":"Quantisierungsbewusstes Training (QAT)","pathname":"/docs/de/blog/quantization-aware-training-qat","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"down-left-and-up-right-to-center","description":"Quantisiere Modelle mit Unsloth und PyTorch auf 4 Bit, um Genauigkeit zurückzugewinnen.","breadcrumbs":[{"label":"Blog"}]},{"id":"ab4316bf164534c25b177f0a9338ead582c3201c","title":"Feinabstimmung von LLMs auf der NVIDIA DGX Station mit Unsloth","pathname":"/docs/de/blog/dgx-station","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"microchip-ai","description":"NVIDIA-DGX-Station-Tutorial zur Feinabstimmung mit Notebooks von Unsloth.","breadcrumbs":[{"label":"Blog"}]},{"id":"3baa81327c4b135f98ea3c328775c251feb9564f","title":"Wie man LLMs mit Unsloth und Docker feinabstimmt","pathname":"/docs/de/blog/how-to-fine-tune-llms-with-unsloth-and-docker","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"docker","description":"Lerne, wie man LLMs feinabstimmt oder Reinforcement Learning (RL) mit dem Docker-Image von Unsloth durchführt.","breadcrumbs":[{"label":"Blog"}]},{"id":"6d9bae5b6da8050c1b9a805e1d9eefc6d4d02f08","title":"Feinabstimmung von LLMs mit NVIDIA DGX Spark und Unsloth","pathname":"/docs/de/blog/fine-tuning-llms-with-nvidia-dgx-spark-and-unsloth","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"sparkle","description":"Tutorial, wie man mit OpenAI gpt-oss auf NVIDIA DGX Spark feinabstimmt und Reinforcement Learning (RL) durchführt.","breadcrumbs":[{"label":"Blog"}]},{"id":"2c09033b89b260a14c0fd8f6283604d65c2ce493","title":"Feinabstimmung von LLMs mit Blackwell, RTX-50-Serie & Unsloth","pathname":"/docs/de/blog/fine-tuning-llms-with-blackwell-rtx-50-series-and-unsloth","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"microchip","description":"Lerne mit unserem Schritt-für-Schritt-Leitfaden, wie man LLMs auf NVIDIAs Blackwell RTX-50-Serie und B200-GPUs feinabstimmt.","breadcrumbs":[{"label":"Blog"}]},{"id":"71a7a6b820be89726c6604431f537ea6caade3ef","title":"Entfessle die Power von AMD: Offizieller Support für Unsloth ist da!","pathname":"/docs/de/blog/entfessle-die-power-von-amd-offizieller-support-fur-unsloth-ist-da","siteSpaceId":"sitesp_L6rLB","lang":"de","icon":"square-up-right","description":"Der Support für AMD-GPUs in Unsloth ist jetzt offiziell. Feinabstimme LLMs bis zu 2x schneller mit ~70 % weniger Speicher, ohne NVIDIA-Hardware.","breadcrumbs":[{"label":"Blog"}]},{"id":"4512843f43e3d51c6fd567ba8d1d41bd61aac871","title":"Documentation Unsloth","pathname":"/docs/fr","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9a5","description":"Unsloth est un framework open source pour exécuter et entraîner des modèles.","breadcrumbs":[{"label":"Commencer"}]},{"id":"fae0ce37756fb39f97798fc8e7e12aa27c33bc88","title":"Affinage pour débutants","pathname":"/docs/fr/commencer/fine-tuning-for-beginners","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"2b50","description":"","breadcrumbs":[{"label":"Commencer"}]},{"id":"47370eb8ad54f7198da9c5711a7d50efdf003784","title":"Configuration requise pour Unsloth","pathname":"/docs/fr/commencer/fine-tuning-for-beginners/unsloth-requirements","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f6e0","description":"Voici les exigences d'Unsloth, y compris les exigences système et de VRAM GPU.","breadcrumbs":[{"label":"Commencer"},{"label":"Affinage pour débutants","emoji":"2b50"}]},{"id":"a4c177d0cc0d485af2566701bc0c3f6f94175629","title":"FAQ + L'affinage est-il fait pour moi ?","pathname":"/docs/fr/commencer/fine-tuning-for-beginners/faq-+-is-fine-tuning-right-for-me","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f914","description":"Si vous hésitez à savoir si l'affinage vous convient, regardez ici ! Découvrez les idées reçues sur l'affinage, comparez-le au RAG et bien plus encore :","breadcrumbs":[{"label":"Commencer"},{"label":"Affinage pour débutants","emoji":"2b50"}]},{"id":"2bdfa5349e5d636154595876b11e5db5e1f9e9d6","title":"Carnets Unsloth","pathname":"/docs/fr/commencer/unsloth-notebooks","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4d2","description":"Carnets d'affinage : explorez le catalogue Unsloth.","breadcrumbs":[{"label":"Commencer"}]},{"id":"ca98e2e30b9d7bfca2555b29e750ac7d5e3e674c","title":"Catalogue de modèles Unsloth","pathname":"/docs/fr/commencer/unsloth-model-catalog","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f52e","description":"","breadcrumbs":[{"label":"Commencer"}]},{"id":"7ffcb5efbe9e4cf09524c2b084bceb779bedad90","title":"Installation d'Unsloth","pathname":"/docs/fr/commencer/install","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4e5","description":"Apprenez à installer Unsloth localement ou en ligne.","breadcrumbs":[{"label":"Commencer"}]},{"id":"017499fb6962e259f964d419ed274c68a2f78f23","title":"Installer Unsloth via pip et uv","pathname":"/docs/fr/commencer/install/pip-install","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"desktop-arrow-down","description":"Pour installer Unsloth localement via pip, suivez les étapes ci-dessous :","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"14c4a3ccd3644f3f4905e3d528cdb66ead97b8ce","title":"Installer Unsloth sur MacOS","pathname":"/docs/fr/commencer/install/mac","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"apple","description":"","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"897f358231280f27c42fdd0fe983b7be0dd75dfc","title":"Comment affiner des LLM sur Windows avec Unsloth (guide étape par étape)","pathname":"/docs/fr/commencer/install/windows-installation","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"windows","description":"Découvrez comment installer Unsloth sur Windows pour commencer à affiner des LLM localement.","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"753bcf3b457799d53f6336913d21ff3056ea8655","title":"Installer Unsloth via Docker","pathname":"/docs/fr/commencer/install/docker","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"docker","description":"Installez Unsloth à l'aide de notre conteneur Docker officiel","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"961323c1e9ab5878c28afd53cf87474039080525","title":"Mettre à jour Unsloth","pathname":"/docs/fr/commencer/install/updating","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"arrow-rotate-right","description":"Pour mettre à jour ou utiliser une ancienne version d'Unsloth, suivez les étapes ci-dessous :","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"4428ac8b919c6d934a1260003dc58ba0547d6273","title":"Guide d'affinage des LLM sur GPU AMD avec Unsloth","pathname":"/docs/fr/commencer/install/amd","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"square-up-right","description":"Apprenez à affiner de grands modèles de langage (LLM) sur des GPU AMD avec Unsloth.","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"b3a0a3e07dbd65d73c215148fac229cf552115ac","title":"Affinage des LLM sur GPU Intel avec Unsloth","pathname":"/docs/fr/commencer/install/intel","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"info","description":"Apprenez à entraîner et à affiner de grands modèles de langage sur des GPU Intel.","breadcrumbs":[{"label":"Commencer"},{"label":"Installation d'Unsloth","emoji":"1f4e5"}]},{"id":"9bbe6f156adffaddead7109d8475ab4a8547be46","title":"Guide d'affinage des LLM","pathname":"/docs/fr/commencer/fine-tuning-llms-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9ec","description":"Apprenez toutes les bases et les meilleures pratiques de l'affinage. Adapté aux débutants.","breadcrumbs":[{"label":"Commencer"}]},{"id":"106dfb3c53d8e4d182108e27a2e91ce2b4f38210","title":"Guide des jeux de données","pathname":"/docs/fr/commencer/fine-tuning-llms-guide/datasets-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4c8","description":"Apprenez à créer et préparer un jeu de données pour l'affinage.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide d'affinage des LLM","emoji":"1f9ec"}]},{"id":"f4254d24125c1e7ad4d5ad30f1a9c8f7d36e7c4d","title":"Guide des hyperparamètres pour l'affinage LoRA","pathname":"/docs/fr/commencer/fine-tuning-llms-guide/lora-hyperparameters-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9e0","description":"Apprenez pas à pas les meilleurs réglages d'affinage des LLM : rang et alpha LoRA, époques, taille de lot + accumulation de gradients, QLoRA vs LoRA, modules cibles, et plus encore.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide d'affinage des LLM","emoji":"1f9ec"}]},{"id":"6a7779ad1adc734a55034f8880a76401f04d84c5","title":"Quel modèle dois-je utiliser pour l'affinage ?","pathname":"/docs/fr/commencer/fine-tuning-llms-guide/what-model-should-i-use","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"2753","description":"","breadcrumbs":[{"label":"Commencer"},{"label":"Guide d'affinage des LLM","emoji":"1f9ec"}]},{"id":"9418759bd903302ac509b6da96465b4bb2e40cfe","title":"Tutoriel : comment affiner Llama-3 et l'utiliser dans Ollama","pathname":"/docs/fr/commencer/fine-tuning-llms-guide/tutorial-how-to-finetune-llama-3-and-use-in-ollama","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f999","description":"Guide du débutant pour créer un assistant personnel personnalisé (comme ChatGPT) à exécuter localement sur Ollama","breadcrumbs":[{"label":"Commencer"},{"label":"Guide d'affinage des LLM","emoji":"1f9ec"}]},{"id":"85e0cbe0d1323be98ebb2569af86994406e2a2d6","title":"Guide de l'apprentissage par renforcement (RL)","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4a1","description":"Découvrez tout sur l'apprentissage par renforcement (RL) et comment entraîner votre propre modèle de raisonnement DeepSeek-R1 avec Unsloth en utilisant GRPO. Un guide complet du débutant au niveau avancé.","breadcrumbs":[{"label":"Commencer"}]},{"id":"93e8b0e7e053d1516b2631829beedee63c88ea4c","title":"Apprentissage par renforcement GRPO avec un contexte 7x plus long","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/grpo-long-context","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f300","description":"Découvrez comment Unsloth permet un affinage RL à contexte ultra long.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"828a94c1637e052b62bda4ce216caa066667b218","title":"Apprentissage par renforcement vision (VLM RL)","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/vision-reinforcement-learning-vlm-rl","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f441-1f5e8","description":"Entraînez des modèles vision/multimodaux via GRPO et RL avec Unsloth !","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"92ec1eda06a9686b604ef754fe5b88c169de1cae","title":"Apprentissage par renforcement en FP8","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/fp8-reinforcement-learning","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f3b1","description":"Entraînez l'apprentissage par renforcement (RL) et GRPO en précision FP8 avec Unsloth.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"3b308de419b284a6a192a8a4ada6d06eeb57ae12","title":"Tutoriel : entraînez votre propre modèle de raisonnement avec GRPO","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/tutorial-train-your-own-reasoning-model-with-grpo","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"26a1","description":"Guide du débutant pour transformer un modèle comme Llama 3.1 (8B) en modèle de raisonnement en utilisant Unsloth et GRPO.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"94ab2d9623c104a6b6a1aa04195e1ef51b45b84c","title":"Documentation avancée sur l'apprentissage par renforcement","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/advanced-rl-documentation","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9e9","description":"Paramètres de documentation avancés lors de l'utilisation d'Unsloth avec GRPO.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"7fd0dddd3d571760e996aecd18e08f8ab58fbf5b","title":"Apprentissage par renforcement GSPO","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/advanced-rl-documentation/gspo-reinforcement-learning","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"lightbulb-on","description":"Entraînez-vous avec le RL GSPO (Group Sequence Policy Optimization) dans Unsloth.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"},{"label":"Documentation avancée sur l'apprentissage par renforcement","emoji":"1f9e9"}]},{"id":"fc22bd16a489581349db130b2b5b1c7d4a9f2195","title":"Hacking des récompenses en RL","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/advanced-rl-documentation/rl-reward-hacking","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"treasure-chest","description":"Découvrez ce qu'est le hacking des récompenses en apprentissage par renforcement et comment le contrer.","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"},{"label":"Documentation avancée sur l'apprentissage par renforcement","emoji":"1f9e9"}]},{"id":"c10868db27b3a6ca9f7465ed5087d86177024eb0","title":"FP16 vs BF16 pour le RL","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/advanced-rl-documentation/fp16-vs-bf16-for-rl","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"2049","description":"L'article « Defeating the Training-Inference Mismatch via FP16 » https://arxiv.org/pdf/2510.26788 montre que l'utilisation du float16 est meilleure que le bfloat16","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"},{"label":"Documentation avancée sur l'apprentissage par renforcement","emoji":"1f9e9"}]},{"id":"2002066c130086ecdd2c6778ee34ab38a981a1dd","title":"RL efficace en mémoire","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/memory-efficient-rl","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"memory","description":"","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"d9644059f502160bc700a9f4d6e0821533b1b59b","title":"Entraînement par optimisation des préférences - DPO, ORPO et KTO","pathname":"/docs/fr/commencer/reinforcement-learning-rl-guide/preference-dpo-orpo-and-kto","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f3c6","description":"Découvrez l'affinage de l'alignement des préférences avec DPO, GRPO, ORPO ou KTO via Unsloth, suivez les étapes ci-dessous :","breadcrumbs":[{"label":"Commencer"},{"label":"Guide de l'apprentissage par renforcement (RL)","emoji":"1f4a1"}]},{"id":"22a56cb154401d43a94d87fa72ce6dfde69b18e3","title":"Présentation d'Unsloth Studio","pathname":"/docs/fr/nouveau/studio","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9a5","description":"Exécutez et entraînez des modèles d'IA localement avec Unsloth Studio.","breadcrumbs":[{"label":"Nouveau"}]},{"id":"a4ab77cc976ec484d13317b14422a754e2f1d4e8","title":"Commencez avec Unsloth Studio","pathname":"/docs/fr/nouveau/studio/start","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"bolt","description":"Un guide pour commencer avec le studio d'affinage, les recettes de données, l'exportation de modèles et le chat.","breadcrumbs":[{"label":"Nouveau"},{"label":"Présentation d'Unsloth Studio","emoji":"1f9a5"}]},{"id":"02de109936ce31121bffae3333822baa85a115f0","title":"Comment exécuter des modèles avec Unsloth Studio","pathname":"/docs/fr/nouveau/studio/chat","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"comment-dots","description":"Exécutez localement des modèles d'IA, des LLM et des GGUF avec Unsloth Studio.","breadcrumbs":[{"label":"Nouveau"},{"label":"Présentation d'Unsloth Studio","emoji":"1f9a5"}]},{"id":"3ca98ab552557cd126d152e81d801c8cb0caa73c","title":"Installation d'Unsloth Studio","pathname":"/docs/fr/nouveau/studio/install","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"arrow-down-to-square","description":"Découvrez comment installer Unsloth Studio sur votre appareil local.","breadcrumbs":[{"label":"Nouveau"},{"label":"Présentation d'Unsloth Studio","emoji":"1f9a5"}]},{"id":"ec921eec2a37820cfd0cf432328d619c9fa1080b","title":"Recettes de données Unsloth","pathname":"/docs/fr/nouveau/studio/data-recipe","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"hat-chef","description":"Découvrez comment créer, construire et modifier des jeux de données avec les recettes de données d'Unsloth Studio.","breadcrumbs":[{"label":"Nouveau"},{"label":"Présentation d'Unsloth Studio","emoji":"1f9a5"}]},{"id":"817a1275219e1e8d86fe100d223ac4b0862ab3a1","title":"Exporter des modèles avec Unsloth Studio","pathname":"/docs/fr/nouveau/studio/export","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"box-isometric","description":"Découvrez comment exporter vos fichiers de modèle safetensor ou LoRA vers GGUF ou d'autres formats.","breadcrumbs":[{"label":"Nouveau"},{"label":"Présentation d'Unsloth Studio","emoji":"1f9a5"}]},{"id":"17e668a82f6382e48c91ce24710a758a8d7eb23c","title":"Mises à jour d'Unsloth","pathname":"/docs/fr/nouveau/changelog","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"sparkles","description":"Journal des modifications d'Unsloth pour nos dernières versions, améliorations et corrections.","breadcrumbs":[{"label":"Nouveau"}]},{"id":"4a2b83ac4bf0233da80a1e3b6ab9fb218108742c","title":"Qwen3.6 - Comment l'exécuter localement","pathname":"/docs/fr/modeles/qwen3.6","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f49c","description":"Exécutez localement les nouveaux modèles Qwen3.6-27B et 35B-A3B !","breadcrumbs":[{"label":"Modèles"}]},{"id":"fa9788be3ba8450d14c81331c0249f2201968a40","title":"Gemma 4 - Comment l'exécuter localement","pathname":"/docs/fr/modeles/gemma-4","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"2728","description":"Exécutez localement les nouveaux modèles Gemma 4 de Google, y compris E2B, E4B, 26B A4B et 31B.","breadcrumbs":[{"label":"Modèles"}]},{"id":"11ad65c8cb780dbffa6556d9554801824345ccfa","title":"Guide d'affinage Gemma 4","pathname":"/docs/fr/modeles/gemma-4/train","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"flask-gear","description":"Entraînez Gemma 4 de Google avec Unsloth.","breadcrumbs":[{"label":"Modèles"},{"label":"Gemma 4 - Comment l'exécuter localement","emoji":"2728"}]},{"id":"84a40d9b3b6f93f936dda7731d24ec92ca78678b","title":"NVIDIA Nemotron 3 Nano Omni - Comment l'exécuter localement","pathname":"/docs/fr/modeles/nemotron-3-nano-omni","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9e9","description":"Exécutez et affinez Nemotron-3-Nano-Omni-30B-A3B localement sur votre appareil !","breadcrumbs":[{"label":"Modèles"}]},{"id":"bc4b3913f03692fd54555ddd8ad4faf9c0654c47","title":"Kimi K2.6 - Comment l'exécuter localement","pathname":"/docs/fr/modeles/kimi-k2.6","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f95d","description":"Guide étape par étape pour exécuter Kimi-K2.6 sur votre propre appareil local.","breadcrumbs":[{"label":"Modèles"}]},{"id":"5dc1787d57509ad87481c06d36ca9df87c35c053","title":"Qwen3.5 - Comment l'exécuter localement","pathname":"/docs/fr/modeles/qwen3.5","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f49c","description":"Exécutez les nouveaux LLM Qwen3.5, y compris Medium : Qwen3.5-35B-A3B, 27B, 122B-A10B, Small : Qwen3.5-0.8B, 2B, 4B, 9B et 397B-A17B sur votre appareil local !","breadcrumbs":[{"label":"Modèles"}]},{"id":"f7f3dcef9bc832cb26935d1dd2773d37a53b1d88","title":"Guide d'affinage Qwen3.5","pathname":"/docs/fr/modeles/qwen3.5/fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"flask-gear","description":"Apprenez à affiner les LLM Qwen3.5 avec Unsloth.","breadcrumbs":[{"label":"Modèles"},{"label":"Qwen3.5 - Comment l'exécuter localement","emoji":"1f49c"}]},{"id":"0ca7fe7921ecd1e2e030c324a794e27e90007ffb","title":"Benchmarks Qwen3.5 GGUF","pathname":"/docs/fr/modeles/qwen3.5/gguf-benchmarks","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"chart-fft","description":"Découvrez les performances des GGUF dynamiques d'Unsloth + une analyse de la perplexité, de la divergence KL et du MXFP4.","breadcrumbs":[{"label":"Modèles"},{"label":"Qwen3.5 - Comment l'exécuter localement","emoji":"1f49c"}]},{"id":"40461af86e39f7eb7ece0eb6f1e6393ccedb5b02","title":"GLM-5.1 - Comment l'exécuter localement","pathname":"/docs/fr/modeles/glm-5.1","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"z","description":"Exécutez le nouveau modèle GLM-5.1 de Z.ai sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"}]},{"id":"41f2e6bbbe116f730cf3549300cf28972e8c72f4","title":"Tutoriels sur les grands modèles de langage (LLM)","pathname":"/docs/fr/modeles/tutorials","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f680","description":"","breadcrumbs":[{"label":"Modèles"}]},{"id":"cdcd2213e0d2a7563e4a8dabe621d0042b1ba8b2","title":"Qwen3 - Comment exécuter et affiner","pathname":"/docs/fr/modeles/tutorials/qwen3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f320","description":"Apprenez à exécuter et à affiner Qwen3 localement avec Unsloth + nos quantifications dynamiques 2.0","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"d45a882f894f981162b0fef555540add4b7d72e1","title":"Qwen3-VL : Guide d'exécution","pathname":"/docs/fr/modeles/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-vl-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f320","description":"Apprenez à affiner et à exécuter Qwen3-VL localement avec Unsloth.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"},{"label":"Qwen3 - Comment exécuter et affiner","emoji":"1f320"}]},{"id":"08b638079f720f03f4ae61f30ea4084f852da73b","title":"Qwen3-2507 : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/qwen3-how-to-run-and-fine-tune/qwen3-2507","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f320","description":"Exécutez localement sur votre appareil les versions Thinking et Instruct de Qwen3-30B-A3B-2507 et 235B-A22B !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"},{"label":"Qwen3 - Comment exécuter et affiner","emoji":"1f320"}]},{"id":"76ae8488a3005b9cbd5dd14d8f7445641a902fff","title":"MiniMax-M2.7 - Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/minimax-m27","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"waveform","description":"Exécutez le LLM MiniMax-M2.7 localement sur votre propre appareil !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"a8d2bea93c712c2574751649afb854b785762586","title":"GLM-5 : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/glm-5","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"z","description":"Exécutez le nouveau modèle GLM-5 de Z.ai sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"54af960e93fa00757e50c6eeafca4d56f706cf90","title":"Kimi K2.5 : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/kimi-k2.5","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f95d","description":"Guide pour exécuter Kimi-K2.5 sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"e093c577dca65e62304722a0d0d3198ab33d3de8","title":"GLM-4.7-Flash : Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/glm-4.7-flash","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"z","description":"Exécutez et affinez GLM-4.7-Flash localement sur votre appareil !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"a6f4f8425254815d780258ea77aa8ed7d94c90b6","title":"MiniMax-M2.5 : Guide d'exécution","pathname":"/docs/fr/modeles/tutorials/minimax-m25","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"waveform","description":"Exécutez MiniMax-M2.5 localement sur votre propre appareil !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"e602d78d631056880787b680897d774bfbdacc01","title":"Qwen3-Coder : Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/qwen3-coder-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f320","description":"Exécutez Qwen3-Coder-30B-A3B-Instruct et 480B-A35B localement avec les quantifications dynamiques d'Unsloth.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"18b45693c88a48fd1d16532bb5dfceb5166df38e","title":"Gemma 3 - Guide d'exécution","pathname":"/docs/fr/modeles/tutorials/gemma-3-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"google","description":"Comment exécuter efficacement Gemma 3 avec nos GGUF sur llama.cpp, Ollama, Open WebUI et comment l'affiner avec Unsloth !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"316806651eee65f5b87bb9c872d0f7ad0026b962","title":"Gemma 3n : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/gemma-3-how-to-run-and-fine-tune/gemma-3n-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"google","description":"Exécutez localement le nouveau Gemma 3n de Google avec des GGUF dynamiques sur llama.cpp, Ollama, Open WebUI et affinez-le avec Unsloth !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"},{"label":"Gemma 3 - Guide d'exécution","icon":"google"}]},{"id":"2033b6a7791704447730e8398ce00576aed8425a","title":"DeepSeek-OCR 2 : Guide d'exécution et d'affinage","pathname":"/docs/fr/modeles/tutorials/deepseek-ocr-2","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f433","description":"Guide pour exécuter et affiner DeepSeek-OCR-2 localement.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"0774814301220b49efb3fccdce0c6102dcad153c","title":"GLM-4.7 : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/glm-4.7","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"z","description":"Un guide sur la façon d'exécuter le modèle GLM-4.7 de Z.ai sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"f02184c12ad2b22efb34111252f7a23753e3f4fc","title":"Comment exécuter Qwen-Image-2512 localement dans ComfyUI","pathname":"/docs/fr/modeles/tutorials/qwen-image-2512","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f49f","description":"Tutoriel étape par étape pour exécuter Qwen-Image-2512 sur votre appareil local avec ComfyUI.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"71ea20aa6692f64a79b8514a198ef941fd0c8632","title":"Tutoriel pour exécuter Qwen-Image-2512 dans stable-diffusion.cpp","pathname":"/docs/fr/modeles/tutorials/qwen-image-2512/stable-diffusion.cpp","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f3a8","description":"Tutoriel pour utiliser Qwen-Image-2512 dans stable-diffusion.cpp.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"},{"label":"Comment exécuter Qwen-Image-2512 localement dans ComfyUI","emoji":"1f49f"}]},{"id":"ceadbfb0f2e489cd5e42c6a03ec971d9d0e0700e","title":"Devstral 2 - Guide d'exécution","pathname":"/docs/fr/modeles/tutorials/devstral-2","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4d9","description":"Guide pour exécuter localement les modèles Mistral Devstral 2 : 123B-Instruct-2512 et Small-2-24B-Instruct-2512.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"b7e8e6530edc0f8d90d1dd7614a8a127f98532da","title":"Ministral 3 - Guide d'exécution","pathname":"/docs/fr/modeles/tutorials/ministral-3","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f431","description":"Guide pour les modèles Mistral Ministral 3, pour les exécuter ou les affiner localement sur votre appareil","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"6c1e05625f15370e1907a3fcd77edd84cccd087f","title":"DeepSeek-OCR : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/deepseek-ocr-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f433","description":"Guide pour exécuter et affiner DeepSeek-OCR localement.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"1be04c8e20275b503c8a5a9caf1ef49be79c8ff1","title":"Kimi K2 Thinking : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/kimi-k2-thinking-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f319","description":"Guide pour exécuter Kimi-K2-Thinking et Kimi-K2 sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"dae16239a066aab3e73d5ba62459a60853a54430","title":"GLM-4.6 : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/glm-4.6-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"z","description":"Un guide sur la façon d'exécuter les modèles Z.ai GLM-4.6 et GLM-4.6V-Flash sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"29445d0c137d738fbbd518145144f5451be337d7","title":"Qwen3-Next : Guide d'exécution locale","pathname":"/docs/fr/modeles/tutorials/qwen3-next","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f320","description":"Exécutez localement sur votre appareil les versions Qwen3-Next-80B-A3B-Instruct et Thinking !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"3c9d2836b7f65414c0a313d922358338d23d3862","title":"FunctionGemma : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/functiongemma","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"google","description":"Apprenez à exécuter et à affiner FunctionGemma localement sur votre appareil et votre téléphone.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"b3b1fa5961974e1d851732430e0f9edd08662c7c","title":"DeepSeek-V3.1 : Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/deepseek-v3.1-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f40b","description":"Un guide sur la façon d'exécuter DeepSeek-V3.1 et Terminus sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"bf22065882c875337e6a9aea8f14f3ab25542607","title":"DeepSeek-R1-0528 : Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/deepseek-r1-0528-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f40b","description":"Un guide sur la façon d'exécuter DeepSeek-R1-0528, y compris Qwen3, sur votre propre appareil local !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"c528ad8519f4ca046cd1328b5bdfae3996e0899b","title":"Liquid LFM2.5 : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/lfm2.5","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4a7","description":"Exécutez et affinez LFM2.5 Instruct et Vision localement sur votre appareil !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"8f26be2f0d48a809bafc8762a3ec56ade9b747af","title":"Magistral : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/magistral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4a5","description":"Découvrez Magistral - les nouveaux modèles de raisonnement de Mistral.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"d97697685cec88b23ee4d6443b03696bf640cb58","title":"IBM Granite 4.0","pathname":"/docs/fr/modeles/tutorials/ibm-granite-4.0","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"cube","description":"Comment exécuter IBM Granite-4.0 avec les GGUF Unsloth sur llama.cpp, Ollama et comment l'affiner !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"fca3c3fee394776398920e2c9e3428f5ec51196c","title":"Llama 4 : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/llama-4-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f999","description":"Comment exécuter Llama 4 localement en utilisant nos GGUF dynamiques, qui récupèrent la précision par rapport à la quantification standard.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"bc5e5c47bb3bbf61a7d7e6e5b5209070925d54ee","title":"Grok 2","pathname":"/docs/fr/modeles/tutorials/grok-2","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"square-x-twitter","description":"Exécutez localement le modèle Grok 2 de xAI !","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"8d1672056da71a0aa082fbd04684a154069c6733","title":"Devstral : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/devstral-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4d9","description":"Exécutez et affinez Mistral Devstral 1.1, y compris Small-2507 et 2505.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"18ff82c8ec22aac46f659c98c536562dad45be0b","title":"Comment exécuter des LLM locaux avec Docker : guide étape par étape","pathname":"/docs/fr/modeles/tutorials/how-to-run-llms-with-docker","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"docker","description":"Apprenez à exécuter des grands modèles de langage (LLM) avec Docker et Unsloth sur votre appareil local.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"dc11579ec5aed1393c09ae05e1a43c23c967f615","title":"DeepSeek-V3-0324 : Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/deepseek-v3-0324-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f433","description":"Comment exécuter DeepSeek-V3-0324 localement en utilisant nos quantifications dynamiques qui récupèrent la précision","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"77b53d22f61fd669ad87b20c49e3612460acea2d","title":"DeepSeek-R1 : Comment l'exécuter localement","pathname":"/docs/fr/modeles/tutorials/deepseek-r1-how-to-run-locally","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f40b","description":"Un guide sur la façon d'exécuter nos quantifications dynamiques 1,58 bits pour DeepSeek-R1 en utilisant llama.cpp.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"7748174ae79b078e6a9c8c4be21b8560134b0eeb","title":"DeepSeek-R1 Dynamic 1.58-bit","pathname":"/docs/fr/modeles/tutorials/deepseek-r1-how-to-run-locally/deepseek-r1-dynamic-1.58-bit","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f433","description":"Consultez les tableaux de comparaison des performances des quantifications GGUF dynamiques d'Unsloth par rapport aux quantifications IMatrix standard.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"},{"label":"DeepSeek-R1 : Comment l'exécuter localement","emoji":"1f40b"}]},{"id":"3a68359f2c51acf87bf9a70a01f9fd0cdf7c5e8a","title":"Phi-4 Reasoning : comment l'exécuter et l'affiner","pathname":"/docs/fr/modeles/tutorials/phi-4-reasoning-how-to-run-and-fine-tune","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"windows","description":"Apprenez à exécuter et à affiner localement les modèles de raisonnement Phi-4 avec Unsloth + nos quantifications dynamiques 2.0","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"3677562dc14186ff4f32a2628a809387e65fd0ea","title":"QwQ-32B : comment l'exécuter efficacement","pathname":"/docs/fr/modeles/tutorials/qwq-32b-how-to-run-effectively","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f320","description":"Comment exécuter efficacement QwQ-32B avec nos correctifs de bogues et sans générations infinies + GGUF.","breadcrumbs":[{"label":"Modèles"},{"label":"Tutoriels sur les grands modèles de langage (LLM)","emoji":"1f680"}]},{"id":"d7ed99d74f1997aa8747da14938bbaee3f09d15b","title":"Comment utiliser Unsloth comme point de terminaison API","pathname":"/docs/fr/notions-de-base/api","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"globe","description":"","breadcrumbs":[{"label":"Notions de base"}]},{"id":"44b6f06033c7dbf3b6521a33337058e295acc604","title":"Inférence et déploiement","pathname":"/docs/fr/notions-de-base/inference-and-deployment","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f5a5","description":"Apprenez à enregistrer votre modèle affiné afin de pouvoir l'exécuter dans votre moteur d'inférence préféré.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"0ce33fc68eed069d43cdcfb76b9793ce71c64c1f","title":"Enregistrement au format GGUF","pathname":"/docs/fr/notions-de-base/inference-and-deployment/saving-to-gguf","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"e11ba1106fa93f45a313adb5372f2081e3ad0b90","title":"Décodage spéculatif","pathname":"/docs/fr/notions-de-base/inference-and-deployment/saving-to-gguf/speculative-decoding","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Décodage spéculatif avec llama-server, llama.cpp, vLLM et plus encore pour une inférence 2x plus rapide","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"},{"label":"Enregistrement au format GGUF"}]},{"id":"682151c53afcf1f6d611eb29ad62b7182b5187ea","title":"Guide de déploiement et d'inférence vLLM","pathname":"/docs/fr/notions-de-base/inference-and-deployment/vllm-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Guide pour enregistrer et déployer des LLM sur vLLM afin de servir des LLM en production","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"1ef02dc5c24ab1305556a9b21ced05fca5ca43d9","title":"Arguments du moteur vLLM","pathname":"/docs/fr/notions-de-base/inference-and-deployment/vllm-guide/vllm-engine-arguments","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"},{"label":"Guide de déploiement et d'inférence vLLM"}]},{"id":"4017cca7d27bd8a0f41fd28d9e47f3f9699549d8","title":"Guide de changement à chaud LoRA","pathname":"/docs/fr/notions-de-base/inference-and-deployment/vllm-guide/lora-hot-swapping-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"},{"label":"Guide de déploiement et d'inférence vLLM"}]},{"id":"169851d8ccb3cd6dc872748a239f3bf944e2cd74","title":"Enregistrement vers Ollama","pathname":"/docs/fr/notions-de-base/inference-and-deployment/saving-to-ollama","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"36cb808bf0f748d077633c7ecbb311cce41e282a","title":"Déploiement de modèles sur LM Studio","pathname":"/docs/fr/notions-de-base/inference-and-deployment/lm-studio","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Enregistrement des modèles en GGUF afin de pouvoir les exécuter et les déployer sur LM Studio","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"4f981fad0586ec63b4986a1a83a7a1dd61f82b94","title":"Comment installer LM Studio CLI dans le terminal Linux","pathname":"/docs/fr/notions-de-base/inference-and-deployment/lm-studio/how-to-install-lm-studio-cli-in-linux-terminal","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f47e","description":"Guide d'installation de LM Studio CLI sans interface utilisateur dans une instance de terminal.","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"},{"label":"Déploiement de modèles sur LM Studio"}]},{"id":"b4083297e9c4dc4c5eedc209c17ef65ddd265e4e","title":"Guide de déploiement et d'inférence SGLang","pathname":"/docs/fr/notions-de-base/inference-and-deployment/sglang-guide","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Guide pour enregistrer et déployer des LLM sur SGLang afin de servir des LLM en production","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"b7833386edaca08d62cb22de0c06676726d89d43","title":"Guide de déploiement llama-server et point de terminaison OpenAI","pathname":"/docs/fr/notions-de-base/inference-and-deployment/llama-server-and-openai-endpoint","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Déploiement via llama-server avec un point de terminaison compatible OpenAI","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"feb774ec9526706cf5885edfba3f100fbb320ea6","title":"Comment exécuter et déployer des LLM sur votre téléphone iOS ou Android","pathname":"/docs/fr/notions-de-base/inference-and-deployment/deploy-llms-phone","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4f1","description":"Tutoriel pour affiner votre propre LLM et le déployer sur votre Android ou iPhone avec ExecuTorch.","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"6dba44c4c4f004bdca413ea55834649bee26efe4","title":"Dépannage de l'inférence","pathname":"/docs/fr/notions-de-base/inference-and-deployment/troubleshooting-inference","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Si vous rencontrez des problèmes lors de l'exécution ou de l'enregistrement de votre modèle.","breadcrumbs":[{"label":"Notions de base"},{"label":"Inférence et déploiement","emoji":"1f5a5"}]},{"id":"6c4a155ae35df476974e25b66af4db620dffaf2c","title":"Comment exécuter des LLM locaux avec Claude Code","pathname":"/docs/fr/notions-de-base/claude-code","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"claude","description":"Guide pour utiliser des modèles ouverts avec Claude Code sur votre appareil local.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"0bb2f0a13e244fd2f0ea640c96c4e297bf83db93","title":"Comment exécuter des LLM locaux avec OpenAI Codex","pathname":"/docs/fr/notions-de-base/codex","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"openai","description":"Utilisez des modèles ouverts avec OpenAI Codex sur votre appareil localement.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"1bdc7fef3d7496e8390edc3622f217d2cc26329f","title":"Affinage multi-GPU avec Unsloth","pathname":"/docs/fr/notions-de-base/multi-gpu-training-with-unsloth","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"rectangle-history","description":"Apprenez à affiner des LLM sur plusieurs GPU et avec parallélisme grâce à Unsloth.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"a09f64a6086d2b9e787e7c01c6d267c6ada0a6a0","title":"Affinage multi-GPU avec Distributed Data Parallel (DDP)","pathname":"/docs/fr/notions-de-base/multi-gpu-training-with-unsloth/ddp","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Apprenez à utiliser l'interface CLI d'Unsloth pour entraîner sur plusieurs GPU avec Distributed Data Parallel (DDP) !","breadcrumbs":[{"label":"Notions de base"},{"label":"Affinage multi-GPU avec Unsloth","icon":"rectangle-history"}]},{"id":"092cf26abaedd940ff2a72e0b51363bc08782323","title":"Guide d'affinage des modèles d'embeddings avec Unsloth","pathname":"/docs/fr/notions-de-base/embedding-finetuning","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f50e","description":"Apprenez à affiner facilement des modèles d'embeddings avec Unsloth.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"658e372c1319ec78ca0371d549bc63e3fdd303d1","title":"Affinez les modèles MoE 12x plus vite avec Unsloth","pathname":"/docs/fr/notions-de-base/faster-moe","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f48e","description":"Entraînez des LLM MoE localement à l'aide du guide Unsloth.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"16606095d5a697b2f0c9e168e53a334ae0d4ca27","title":"Guide d'affinage de la synthèse vocale (TTS)","pathname":"/docs/fr/notions-de-base/text-to-speech-tts-fine-tuning","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f50a","description":"Découvrez comment affiner des modèles vocaux TTS et STT avec Unsloth.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"8e3d570b34072053937ce45a7a2125f403689ec3","title":"GGUF dynamiques 2.0 d'Unsloth","pathname":"/docs/fr/notions-de-base/unsloth-dynamic-2.0-ggufs","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9a5","description":"Une grande nouvelle amélioration de nos quantifications dynamiques !","breadcrumbs":[{"label":"Notions de base"}]},{"id":"350e0b309b3d2e742a0ce946a613e74c5e82d980","title":"GGUF dynamiques d'Unsloth sur Aider Polyglot","pathname":"/docs/fr/notions-de-base/unsloth-dynamic-2.0-ggufs/unsloth-dynamic-ggufs-on-aider-polyglot","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f9a5","description":"Performances des GGUF dynamiques d'Unsloth sur les benchmarks Aider Polyglot","breadcrumbs":[{"label":"Notions de base"},{"label":"GGUF dynamiques 2.0 d'Unsloth","emoji":"1f9a5"}]},{"id":"f47c3320cb986c4fe011958fe46fbbdef0e37e35","title":"Guide d'appel d'outils pour les LLM locaux","pathname":"/docs/fr/notions-de-base/tool-calling-guide-for-local-llms","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"screwdriver-wrench","description":"","breadcrumbs":[{"label":"Notions de base"}]},{"id":"d5d9395669b4b71402e8f13b4b75e8385568af1e","title":"Affinage vision","pathname":"/docs/fr/notions-de-base/vision-fine-tuning","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f441","description":"Apprenez à affiner des LLM vision/multimodaux avec Unsloth","breadcrumbs":[{"label":"Notions de base"}]},{"id":"80484284c3437da1a5e2fc9cf043834fffd2943e","title":"Dépannage et FAQ","pathname":"/docs/fr/notions-de-base/troubleshooting-and-faqs","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"26a0","description":"Conseils pour résoudre les problèmes et questions fréquemment posées.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"a198cfc28f608fe1b7c87d4b263f987c276c6a81","title":"Hugging Face Hub, débogage XET","pathname":"/docs/fr/notions-de-base/troubleshooting-and-faqs/hugging-face-hub-xet-debugging","siteSpaceId":"sitesp_TGKTc","lang":"fr","description":"Débogage, résolution des téléchargements bloqués, coincés et lents","breadcrumbs":[{"label":"Notions de base"},{"label":"Dépannage et FAQ","emoji":"26a0"}]},{"id":"62fe9263ff68a062df2ebefa3b02ed09c677ebb4","title":"Modèles de chat","pathname":"/docs/fr/notions-de-base/chat-templates","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4ac","description":"Découvrez les fondements et les options de personnalisation des modèles de chat, y compris les formats Conversational, ChatML, ShareGPT, Alpaca, et plus encore !","breadcrumbs":[{"label":"Notions de base"}]},{"id":"3e2629b7af8e46722a552d41673baae8c9e21d35","title":"Indicateurs d'environnement Unsloth","pathname":"/docs/fr/notions-de-base/unsloth-environment-flags","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f6e0","description":"Indicateurs avancés pouvant être utiles si vos affinages échouent ou si vous souhaitez désactiver certaines fonctionnalités.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"7e94174930795ab1dc771195c50c516628ce2655","title":"Pré-entraînement continu","pathname":"/docs/fr/notions-de-base/continued-pretraining","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"267b","description":"Aussi appelé affinage continu. Unsloth vous permet de pré-entraîner en continu afin qu'un modèle puisse apprendre une nouvelle langue.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"be565689d8cadaf11a6d552e700eb2a0e7fd3f16","title":"Affinage à partir du dernier checkpoint","pathname":"/docs/fr/notions-de-base/finetuning-from-last-checkpoint","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f3c1","description":"Le checkpointing vous permet d'enregistrer la progression de votre affinage afin de pouvoir le mettre en pause puis le reprendre.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"d748fef4faffef5d30e5a815225fd0b8ce113c7f","title":"Benchmarks Unsloth","pathname":"/docs/fr/notions-de-base/unsloth-benchmarks","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"1f4ca","description":"Benchmarks enregistrés par Unsloth sur les GPU NVIDIA.","breadcrumbs":[{"label":"Notions de base"}]},{"id":"5ed7fe5ad328b9ea2064f629be46409a5616379c","title":"Comment exécuter des modèles d'IA locaux avec OpenCode","pathname":"/docs/fr/integrations/opencode","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"rectangle-vertical","description":"Guide pour connecter des LLM ouverts à OpenCode sur votre appareil local.","breadcrumbs":[{"label":"Intégrations"}]},{"id":"fc0726cf8f72aecd0706033ae732e8e31cb48fcf","title":"Comment exécuter des modèles d'IA locaux avec OpenClaw","pathname":"/docs/fr/integrations/openclaw","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"lobster","description":"Guide pour exécuter des LLM locaux avec OpenClaw.","breadcrumbs":[{"label":"Intégrations"}]},{"id":"b6a49cae84f0f86f978aafa589d0d905608eb666","title":"Comment exécuter des modèles d'IA locaux avec Hermes Agent","pathname":"/docs/fr/integrations/hermes-agent","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"caduceus","description":"Guide d'utilisation des LLM ouverts avec Hermes Agent en local.","breadcrumbs":[{"label":"Intégrations"}]},{"id":"46b648652a387a143566a8cd8b635a0b50c00204","title":"Connecter le SDK Python à Unsloth","pathname":"/docs/fr/integrations/connecter-le-sdk-python-a-unsloth","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"python","description":"Guide pour appeler l'API locale d'Unsloth depuis Python en utilisant les SDK officiels OpenAI ou Anthropic, incluant le streaming, la vision, l'appel de fonctions et les outils côté serveur intégrés d'Unsloth.","breadcrumbs":[{"label":"Intégrations"}]},{"id":"ecc958b339a16f14a987b979d96c3f7d2c0086d1","title":"Connecter Curl et HTTP à Unsloth","pathname":"/docs/fr/integrations/connecter-curl-et-http-a-unsloth","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"spiral","description":"Guide pour interroger l'API d'Unsloth avec curl (ou n'importe quel client HTTP), avec des recettes prêtes à copier-coller pour chaque point de terminaison et fonctionnalité..","breadcrumbs":[{"label":"Intégrations"}]},{"id":"3da83f3b5b7b14de7eeb51382389ba36f855fca1","title":"Entraînement des LLM 3x plus rapide avec les kernels Unsloth + le packing","pathname":"/docs/fr/blog/3x-faster-training-packing","siteSpaceId":"sitesp_TGKTc","lang":"fr","emoji":"26a1","description":"Découvrez comment Unsloth augmente le débit d'entraînement et élimine le gaspillage lié au padding pour l'affinage.","breadcrumbs":[{"label":"Blog"}]},{"id":"4a5176a47bc9d614733d5b512a5f28361c0e8c76","title":"Affinage avec une longueur de contexte de 500K","pathname":"/docs/fr/blog/500k-context-length-fine-tuning","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"ruler-combined","description":"Découvrez comment activer l'affinage d'une fenêtre de contexte de plus de 500K jetons avec Unsloth.","breadcrumbs":[{"label":"Blog"}]},{"id":"e2c1707a97ba45cd1c604e94446a98e276e15828","title":"Entraînement conscient de la quantification (QAT)","pathname":"/docs/fr/blog/quantization-aware-training-qat","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"down-left-and-up-right-to-center","description":"Quantifiez des modèles en 4 bits avec Unsloth et PyTorch pour récupérer la précision.","breadcrumbs":[{"label":"Blog"}]},{"id":"4cf0bfc3d59f854858a4fffec4eab50d6dcc52ef","title":"Affinage des LLM sur NVIDIA DGX Station avec Unsloth","pathname":"/docs/fr/blog/dgx-station","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"microchip-ai","description":"Tutoriel NVIDIA DGX Station sur la façon d'affiner avec les notebooks d'Unsloth.","breadcrumbs":[{"label":"Blog"}]},{"id":"1cb27eb4fb55038e3ce0244a0b0ab5242959a416","title":"Comment affiner des LLM avec Unsloth et Docker","pathname":"/docs/fr/blog/how-to-fine-tune-llms-with-unsloth-and-docker","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"docker","description":"Apprenez à affiner des LLM ou à faire de l'apprentissage par renforcement (RL) avec l'image Docker d'Unsloth.","breadcrumbs":[{"label":"Blog"}]},{"id":"6b63496de50aea16b20eb35d7ed91459bc683484","title":"Affinage des LLM avec NVIDIA DGX Spark et Unsloth","pathname":"/docs/fr/blog/fine-tuning-llms-with-nvidia-dgx-spark-and-unsloth","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"sparkle","description":"Tutoriel sur la façon d'affiner et de faire de l'apprentissage par renforcement (RL) avec OpenAI gpt-oss sur NVIDIA DGX Spark.","breadcrumbs":[{"label":"Blog"}]},{"id":"fe305725c6722e64d44672e8c15ea8ebdabed30c","title":"Affinage des LLM avec Blackwell, la série RTX 50 et Unsloth","pathname":"/docs/fr/blog/fine-tuning-llms-with-blackwell-rtx-50-series-and-unsloth","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"microchip","description":"Apprenez à affiner des LLM sur les GPU Blackwell RTX 50 et B200 de NVIDIA grâce à notre guide étape par étape.","breadcrumbs":[{"label":"Blog"}]},{"id":"83464efd106c354750723164056453cac7cc524f","title":"Libérez la puissance d'AMD : la prise en charge officielle d'Unsloth est arrivée !","pathname":"/docs/fr/blog/liberez-la-puissance-damd-la-prise-en-charge-officielle-dunsloth-est-arrivee","siteSpaceId":"sitesp_TGKTc","lang":"fr","icon":"square-up-right","description":"La prise en charge d'Unsloth pour les GPU AMD est désormais officielle. Affinez des LLM jusqu'à 2x plus rapidement avec environ 70 % de mémoire en moins, sans matériel NVIDIA requis.","breadcrumbs":[{"label":"Blog"}]}]}